Skip to main content

Full text of "soil analysis"

See other formats

Edited by 
G. Schroth and F.L Sinclair 


M — CABI MfeM* 

Trees, Crops and Soil Fertility 
Concepts and Research Methods 

Trees, Crops and Soil Fertility 
Concepts and Research Methods 

Edited by 

G. Schroth 

Biological Dynamics of Forest Fragments Project 
National Institute for Research in the Amazon 


EL. Sinclair 

School of Agricultural and Forest Sciences 

University of Wales 



CABI Publishing 

CABI Publishing is a division of CAB International 

CAB I Publishing 
CAB International 
OxonOXlO 8DE 

CABI Publishing 

44 Brattle Street 

4th Floor 

Cambridge, MA 02138 


Tel: +44(0)1491 832111 
Fax: +44 (0)1491 833508 

Tel: +1 617 395 4056 

Fax: +1 617 354 6875 


©CAB International 2003. All rights reserved. No part of this publication may be 
reproduced in any form or by any means, electronically, mechanically, by photocopying, 
recording or otherwise, without the prior permission of the copyright owners. 

A catalogue record for this book is available from the British Library, London, UK. 

Library of Congress Cataloging-in-Publication Data 

Trees, crops and soil fertility: concepts and research methods 
edited by G. Schroth and F. L. Sinclair. 

p. cm. 
Includes bibliographical references (p. ). 
ISBN 0-85199-593-4 (alk. paper) 

1. Soil fertility. 2. Agroforestry. 3. Soil fertility— Research. 4. Agroforestry— Research. I. 
Schroth, G. (Goetz) II. Sinclair, Fergus L. 
S596.7.T74 2002 


ISBN 85199 593 4 

Typeset by Wyvern 21 Ltd, Bristol 

Printed and bound in the UK by Cromwell Press, Trowbridge 


Contributors ix 

Foreword xi 

M. Swift 

1. Impacts of Trees on the Fertility of Agricultural Soils 1 

G. Schroth and EL. Sinclair 

1.1 Trees and the Development of Agriculture 1 

1.2 Objectives and Structure of the Book 9 

2. Economic Aspects of Soil Fertility Management and 13 
Agroforestry Practices 

A. -M.N. Izac 

2.1 Introduction 13 

2.2 Factors Influencing Farmers' Decisions About Soil 14 
Fertility Management Practices 

2.3 Hierarchy of Agricultural Systems as a Background to 18 
the Understanding of Farmers' Constraints 

2.4 Anatomy of a Decision at the Farm Scale and Economic 18 
Methods for Understanding such Decisions 

2.5 Landscape and Global Scales: Soil Fertility and 32 
Agroforestry Trees as Part of Natural Capital 

3. Designing Experiments and Analysing Data 39 

R. Coe, B. Huwe and G. Schroth 

3.1 Synopsis 39 

/i Contents 

3.2 Experimental Objectives, Treatments and Layout 40 

3.3 Fallow Experiments 48 

3.4 Measurements and Sampling Designs 52 

3.5 Analysing the Data 61 

3.6 Spatial Structure and Its Analysis 67 

4. Soil Organic Matter 77 

G. Schroth, B. Vanlauwe and J. Lehmann 

4.1 Synopsis 77 

4.2 Methods for Total Soil Organic Carbon 86 

4.3 Physical Fractionation Methods 86 

4.4 Chemical Methods 88 

4.5 Biological Methods 89 

5. Soil Nutrient Availability and Acidity 93 

G. Schroth, J. Lehmann and E. Barrios 

5.1 Synopsis 93 

5.2 Methods for Soil Nitrogen 104 

5.3 Methods for Soil Phosphorus 112 

5.4 Methods for Soil Sulphur 121 

5.5 Methods for Potassium, Calcium and Magnesium in Soil 125 

5.6 Methods for Soil Acidity 127 

6. Decomposition and Nutrient Supply from Biomass 131 

G. Schroth 

6.1 Synopsis 131 

6.2 Methods for Biomass and Nutrient Input with Litter 140 

6.3 Methods for Decomposition and Nutrient Release from 143 

6.4 Measures of Resource Quality 148 

7. Nutrient Leaching 151 

J. Lehmann and G. Schroth 

7.1 Synopsis 151 

7.2 Methods for Soil Solution Composition 158 

7.3 Tracer Methods for Nutrient Leaching 163 

7.4 Dyes as Tracers for Preferential Flow Paths 165 

8. Nutrient Capture 167 

G. Schroth and J. Lehmann 

8.1 Synopsis 167 

8.2 Tracer Methods for Nutrient Uptake 174 

Contents vii 

9. Nutrient Exchange with the Atmosphere 181 

G. Schroth and J. Burkhardt 

9.1 Synopsis 181 

9.2 Methods for Atmospheric Nutrient Inputs 185 

9.3 Methods for Nutrient Losses from Burning 188 

10. Soil Structure 191 

M. Grimaldi, G. Schroth, W.G. Teixeira and B. Huwe 

10.1 Synopsis 191 

10.2 Methods for Soil Bulk Density 195 

10.3 Methods for Aggregate Stability 196 

10.4 Methods for Soil Porosity and Pore Size Distribution 198 

10.5 Measuring the Role of Soil Organic Matter in Aggregate 204 

10.6 Soil Micromorphology and Image Analysis 207 

11. Soil Water 209 

W.G. Teixeira, EL. Sinclair, B. Huwe and G. Schroth 

11.1 Synopsis 209 

11.2 Methods for Soil Water Content 215 

11.3 Methods for Soil Water Potential 22 1 

11.4 Methods for Soil Hydraulic Properties 225 

11.5 Estimating Topsoil and Subsoil Water Use with Stable 232 

12. Root Systems 235 

G. Schroth 

12.1 Synopsis 235 

12.2 Methods for Studying Root Distribution 240 

12.3 Distinction of Roots of Different Species 245 

12.4 Methods for Root Dynamics, Production and Turnover 246 

13. Biological Nitrogen Fixation 259 

K.E. Giller 

13.1 Synopsis 259 

13.2 Microbiological Methods for Studying Rhizobia 265 

13.3 Simple Methods for Determining Whether a Legume 265 
is Fixing Nitrogen 

13.4 Isotope-based Methods for Measurement of Nitrogen 266 

13.5 Estimating Nitrogen Fixation in Field Settings 268 

13.6 Methods Based on Nitrogen Fixation Transport 269 

13.7 Estimating Total Amounts of Nitrogen Fixation 270 

viii Contents 

14. Mycorrhizas 271 

D.L. Godbold and R. Sharrock 

14.1 Synopsis 271 

14.2 Inoculation Methods 280 

14.3 Sampling of Plant Roots for Mycorrhizal Studies 281 

14.4 Quantification of Mycorrhizas 281 

14.5 Identification of Mycorrhizal Fungi 283 

14.6 Estimation of Mineral Nutrient Uptake Through 286 

15. Rhizosphere Processes 289 

D. Jones 

15.1 Synopsis 289 

15.2 Obtaining a Representative Sample of Rhizosphere Soil 294 

15.3 Methods for Rhizosphere Soil Chemistry 297 

15.4 Methods for Rhizosphere Biological Activity 298 

15.5 Quantification of Root Carbon Loss into the 300 

15.6 Rhizosphere Mathematical Modelling 300 

16. Soil Macrofauna 303 

P. Lavelle, B. Senapati and E. Barros 

16.1 Synopsis 303 

16.2 Sampling of Macrofauna: the TSBF Methodology 318 

16.3 Other Sampling Methods 320 

16.4 Manipulative Experiments 322 

17. Soil Erosion 325 

M.A. McDonald, A. Lawrence and P.K. Shrestha 

17 A Synopsis 325 

17.2 Quantitative Methods 336 

17.3 Qualitative Methods 341 

References 345 

Index 423 


Edmundo Barrios, Centro International de Agricultura Tropical (CIAT), 

AA 6713 Call, Colombia 

Eleusa Barros, National Institute for Research in the Amazon (INPA), CP 478, 

69011-970 Manaus, AM, Brazil 

Jiirgen Burkhardt, Institute of Agricultural Chemistry, University of Bonn, 

Karlrobert-Kreiten-Str. 13, 53115 Bonn, Germany 

Richard Coe, World Agroforestry Centre, International Centre for Research in 

Agroforestry (ICRAF), PO Box 30677, Nairobi, Kenya 

Ken E. Giller, Plant Production Systems, Department of Plant Sciences, 

Wageningen University, PO Box 430, 6700 AK Wageningen, The Netherlands 

Douglas L. Godbold, School of Agricultural and Forest Sciences, University of 

Wales, Bangor, Gwynedd LL57 2UW, UK 

Michel Grimaldi, Institut de Recherche pour le Developpement (IRD), UR 137, 

Centre de Recherche d'lle de France, 32 ave Henri Varagnat, 93143 Bondy Cedex, 


Bernd Huwe, Institute of Soil Science and Soil Geography, University ofBayreuth, 

95440 Bayreuth, Germany 

Anne-Marie Izac, World Agroforestry Centre, International Centre for Research 

in Agroforestry (ICRAF), PO Box 30677, Nairobi, Kenya 

Davey Jones, School of Agricultural and Forest Sciences, University of Wales, 

Bangor, Gwynedd LL57 2UW, UK 

Patrick Lavelle, Laboratoire d'Ecologie des Sols Tropicaux, 32 ave Henri 

Varagnat, 93143 Bondy Cedex, France 

Anna Lawrence, Environmental Change Institute, University of Oxford, 5 South 

Parks Road, Oxford 0X1 3UB, UK 


Johannes Lehmann, College of Agriculture and Life Sciences, Department of 

Crop and Soil Sciences, Cornell University, 909 Bradfield Hall, Ithaca, NY 14853, 


Morag A. McDonald, School of Agricultural and Forest Sciences, University of 

Wales, Bangor, Gwynedd LL57 2UW, UK 

Gotz Schroth, Biological Dynamics of Forest Fragments Project, National Institute 

for Research in the Amazon (INPA), CP 478, 69011-970 Manaus, AM, Brazil 

Bikram Senapati, Ecology Section, School of Life Sciences, Sambalpur University, 

Jyoti Vihar, 768019, Orissa State, India 

Ruth Sharrock, School of Agricultural and Forest Sciences, University of Wales, 

Bangor, Gwynedd LL57 2UW, UK 

Pratap Kumar Shrestha, Local Initiatives for Biodiversity, Research and 

Development (LI-BIRD), PO Box 324, Bastolathar, Mahendrapool, Pokhara, 


Fergus L. Sinclair, School of Agricultural and Forest Sciences, University of 

Wales, Bangor, Gwynedd LL57 2UW, UK 

Wenceslau G. Teixeira, Empresa Brasileira de Pesquisa Agropecuaria- Amazonia 

Ocidental, CP 319, 69011-970 Manaus, AM, Brazil 

Bernard Vanlauwe, Tropical Soil Biology and Fertility Programme, PO Box 

30592, Nairobi, Kenya 


Soil science benefits from the availability of a wide array of practicable 
methods. There are a number of very useful compendia in which sets of 
these methods are collected together to provide easy reference for the 
intending practitioner. The need for yet another handbook might 
therefore be questioned. This book, however, fulfils several needs that are 
not met in previous volumes. 

First and foremost it is unique in its format and purpose - being not a 
compendium of protocols but a reasoned discussion of the value and utility 
of different methods. Major advances in science are dependent on, and 
sometimes even driven by, the availability of suitable methods by which key 
questions or hypotheses may be answered. Progress may be said to be a 
product of the match between developments in concept and those in 
technique. The structure and content of this book are designed to review 
the ways in which current thinking in terms of the major problems of soil 
fertility can be methodologically attacked. As such the book should be useful 
not only in helping to solve problems but also in provoking new questions. 

The book is written within a particular context - soil fertility 
development under agroforestry. At first this may seem very specific and 
thus limited in appeal and application. But over the last decade or so 
agroforestry research has been one of the most influential in developing 
new insights into soil biology and fertility and thus provides a very suitable 
framework for review of progress. Furthermore, the influence of trees on 
soil is profound and of significance beyond agroforestry systems, so the 
book is likely to be of interest in the wider spheres of agriculture, forestry 
and ecological sciences. 

xii Foreword 

The book covers a spectrum of soil methods broader than most soil 
science handbooks, combining the techniques of soil chemistry and physics 
with a wide scope of biological approaches and aspects of social science. 
This is reflective of the ways in which soil research has moved beyond its 
strict reductionist paradigm to embrace a more holistic and ecosystematic 

The book comes at an appropriate time to review the progress of both 
concept and method in the slow march towards an integrated approach 
to land management. It should not only serve the purpose of directing 
enquiring researchers towards the useful approaches but also act as a 
stimulus for the next wave of soil fertility research. 

Mike Swift 
TSBF, Nairobi, Kenya 

Editors' Note 

This book was produced as part of a project of the International Union of 
Forest Research Organizations (IUFRO) on the development of manuals 
for research in agroforestry in association with the World Agroforestry 
Centre (ICRAF) and the Tropical Agricultural Research and Higher 
Education Centre (CATIE). Mike Swift, former Director of the Tropical 
Soil Biology and Fertility Programme (TSBF), has endorsed the book on 
behalf of the TSBF programme within the International Centre for 
Tropical Agriculture (CIAT). 

Gotz Schroth received funding from the German Ministry of 
Education and Research through the German-Brazilian SHIFT 
programme and from the Brazilian National Council for Scientific and 
Technological Development (CNPq) while working on the book. 

The text was greatly improved by critical comments from several 
colleagues who read the whole manuscript or sections of it, especially Mike 
Swift, Ken E. Giller, Richard Coe, Michel Grimaldi, Bernd Huwe and 
Johannes Lehmann. 

Chapter 1 

Impacts of Trees on the Fertility of 

Agricultural Soils 


biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil; 2 School of Agricultural and Forest Sciences, University of 
Wales, Bangor, Cwynedd LL57 2UW, UK 

1.1 Trees and the Development of Agriculture 

Trees are a natural component of most tropical landscapes, with the 
exception of very dry areas, tropical alpine ecosystems and a few other 
regions with extreme soil or climatic conditions which do not permit the 
establishment of woody perennial plants. In the humid tropics, trees are 
the dominant components of the natural vegetation, the tropical rain 
forests. With decreasing total rainfall and increasing length of the dry 
season, these are replaced by drier forest types and then by savannas. 
Although the density of trees decreases and their crown cover becomes 
more open with increasing aridity of the climate, trees are present and 
play an important role in natural ecosystems from the perhumid tropics 
almost to the fringe of the desert. 

Humans have had a pronounced influence on tropical ecosystems and 
their tree cover for a long time and continue to do so now. Fire and the 
axe, or its modern equivalent the chainsaw, are the main tools employed 
by farmers, planters and pastoralists to reduce tree cover and increase the 
availability of light and soil resources for their crops and pastures. Either 
tropical forests and savannas have been transformed by shifting cultivators 
into patchworks of swidden fields and different stages of fallow regrowth, 
or sedentary farmers have converted natural vegetation more permanently 
into crop fields, pastures, tree crop plantations or human settlements. Some 
of these areas may eventually be abandoned and revert to secondary forest 
or savanna vegetation, unless woody regrowth is prevented by recurrent 
fires, as in the Imperata grasslands of South-east Asia (Garrity et ai, 1997). 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 1 

F.L. Sinclair) 

C. Schroth and F.L. Sinclair 

Although the rise in human populations has caused pronounced 
reductions in tree cover, trees remain an important element of most 
human-dominated landscapes throughout the tropics. Trees provide a 
wide range of important products and services that people in the tropics 
want and need. These range from firewood and construction materials, 
through many different fruits, nuts, medicines, gums, resins and fodder, 
to services such as shade, wind protection and aesthetic and spiritual value 
(Scherr, 1995). Tree cover also provides important habitat for both the 
conservation of wildlife and the utilization of many non-timber forest 
products that people harvest, including a number of valuable plants, fungi 
and game animals. Less visibly, but no less importantly, trees play a crucial 
role in maintaining and regenerating soil fertility through the action of 
their roots and litter. 

Tropical farmers are conscious of these different functions of trees and 
have protected, planted, selected and domesticated trees for thousands of 
years. Shifting cultivation systems in tropical forests depend on the 
regeneration of soil fertility under the forest fallow, which also provides 
game and a variety of products for collection. Shifting cultivators in 
Indonesia and other tropical regions have introduced tree crops such as 
rubber into their swiddens and have created secondary forests enriched 
with valuable trees (Gouyon et at., 1993). Farmers in West African savannas 
maintain valuable trees, which also resist periodical fires, in and around 
their fields, giving rise to a distinct, park-like landscape (Boffa, 1999). 
Planters of tree crops such as cocoa, coffee and tea maintain or establish 
shade trees to reduce pest and disease pressures and nutrient 
requirements of their crops and protect them from climatic extremes (Beer 
et ai, 1998). Since many tree species can fulfil these functions, they may 
choose species that fix atmospheric nitrogen and produce large quantities 
of nutrient- rich litter and prunings, or valuable timber or fruit tree species, 
or any combination of these. Pastoralists value trees for the high nutritional 
value of the fodder from their leaves and fruits, which in seasonally dry 
pastures are still available when the grasses have dried out (Cajas-Giron 
and Sinclair, 2001). They may also prefer to plant trees as living fence posts 
rather than to replace timber posts every few years when they have been 
consumed by termites and fungi. Farmers in eastern and southern Africa 
have recently started to use short-term, planted fallows with legume trees 
to regenerate the fertility of their soils more rapidly than with natural 
fallows and to substitute for mineral nitrogen fertilizer, which is often too 
expensive for them to purchase (Kwesiga et ai, 1999). These fallows may 
also produce valuable animal fodder. 

All these and many more practices that involve growing trees in some 
form of spatial or temporal combination with crops or pastures are known 
as agroforestry. They have drawn substantial interest from scientists and 
development agencies during recent decades, recognizing the fact that 

npacts of Trees on Fertility 

trees can play an important role in income generation and food and fuel 
security for resource-poor rural households, while underpinning the 
sustainability of their farming systems (Cooper et at., 1996). This recent 
upsurge in interest in agroforestry may give the impression that it is a new 
science. This is not the case, as interactions between tree crops and shade 
trees, for example, have been studied by agronomists for more than 100 
years and concepts of nutrient cycling, which are still relevant, were 
developed early on (see, for example, Lock (1888) on shade trees for 
coffee in Sri Lanka). Agroforestry is thus a relatively recent word for a 
much older science and a very old practice. However, efforts to make 
better use of trees in rural development had (and sometimes still 
have) to overcome initial antipathy between agriculture and forestry, 
institutionalized in government departments, research centres and 
educational establishments, which have led to an arbitrary separation of 
research and administration of forests and farming. 

What is agroforestry? 

There are numerous definitions of agroforestry which stress different 
aspects of and expectations about the integration of trees in farming 
landscapes (see, for example, Huxley, 1999). Following the predominant 
definition over the past two decades, 

agroforestry is a set of land use practices that involve the deliberate 
combination of woody perennials including trees, shrubs, palms and bamboos, 
with agricultural crops and/or animals on the same land management unit in 
some form of spatial arrangement or temporal sequence such that there are 
significant ecological and economic interactions among the woody and non- 
woody components. 

(Sinclair, 1999) 

Traditionally, the focus of agroforestry research has been on 
interactions between trees and other components of a system, such as 
crops, soil and climatic factors, on the scale of an individual field or a small 
section of a landscape. This is gradually changing as better understanding 
of small-scale processes enables researchers to scale up their results, and 
as the functions of tree cover manifest at landscape, regional and global 
scales, as a result of larger-scale patterns and processes, become the focus 
of research interest. These functions include water and nutrient cycling 
on the catchment scale, carbon dynamics in the soil-plant-atmosphere 
system and biodiversity. Also, agroforestry practices do not exist in 
isolation, but interact with other land uses across landscapes. A farmer 
maintaining a forest garden or shaded tree crop plantation may also have 
swidden or irrigated rice fields and pasture which occur together within 


Trees and crops 


Trees and pasture 

Predominant land-use objective Type of tree cover 

. Trees on cropland - 2 

Plantation tree crops -2.1.1/2.2.3 


Crops on tree-land ■ 

■ Trees on pasture/range land - 2 

— Pasture on tree-land 

Plantation forest/trees - 
Natural forest/tree cover 

r Plantation tree crops 
Plantation forest/trees 

j- Cropping phase of taungya -2.1 
L Orchards/tree gardens — 2.1 

Forest gardens - 2.1 .2 

Cropping phase of shifting cultivation — 2. 1.2 

Forest/plantation grazing -2.1.1 / 2.2.3 

L Natural forest/tree cover 

Trees and animals 
(not pasture) 

i — Trees as fodder - 

Trees and poultry 
Trees and pigs 

Trees and fish 

1 — Trees and insects 

. Multipurpose utilization of 
natural forest/tree cover 

Trees and tree crops 

Trees on tree-crop land - 2 

Fodder banks -2.2.5 

Silk production 

. Fallow phase of shifting cultivation Improved fallow 

' Primary or secondary forest 

Hunting/collecting non- 
timber forest products 

Fig. 1.1. Primary classification of agroforestry practices based on components/ predominant land-use objective and the type of tree 
cover. Italics indicate examples of agroforestry practices/ numbers refer to the place in Fig. 1 .2 where classification continues (adapted 
from Sinclair/ 1999). 

ipacts of Trees on Fertility 

the same landscape and influence the characteristics of this landscape. This 
wider focus of agroforestry research is reflected in a scale-neutral definition 
of agroforestry as simply 'where trees and agriculture interact' (Sinclair, 

Within this wider view of agroforestry, the landscape scale is emerging 
as a critical unit of analysis (Sinclair, 2001). This is the scale at which 
ecological processes such as the presence and dispersal of fauna and flora, 
water and nutrient flows, microclimate, and pest and disease dynamics are 
significantly influenced by trees. In many fragmented landscapes trees on 
farms, including those shading tree crops or that occur as remnants in 
crop fields or pastures or in riparian corridors, provide key elements of 
the tree cover that determine landscape characteristics. Strategic placing 
of trees in the landscape may prevent, enhance or direct flows of soil, 
water, nutrients, fire and organisms across landscapes (van Noordwijk et 
al., 1999). Where there is a mosaic of agriculture and forest, interactions 
between these land uses determine such important environmental 
functions as the water yield of catchments and landscape-scale biodiversity 

I 2.1 Dispersed 

trees in fields/on range 

2.2 Zoned ■ 

2.1.1 Regular 

- 2.1.2 Irregular 

Live fences 

f- 2.2.1 On boundaries 1— Boundary planting 

Windbreaks/timber belts 

- 2.2.2 On contours 


2.2.3 In rows 

On terraces 

On bunds 

As barrier hedges 

(contour hedgerows) 

— Hedgerow inter- 
cropping/alley cropping 

- 2.2.4 Clumped 

I— 2.2.5 In blocks 

— Production hedges 

I — Fodder banks 

— Farm woodlots 

Fig. 1.2. Secondary classification of agroforestry practices based on density and 
arrangment of the tree component (adapted from Sinclair, 1 999). 

C. Schroth and F.L. Sinclair 

(Guindon, 1996; Bruijnzeel, 1997). In the future, efforts to understand 
and develop the role of trees on farms will increasingly focus on 

In order to facilitate communication, the many ways in which farmers 
use trees within their farming systems have been classified into several 
major types of practice, on the basis of the components that are involved, 
the type of land on which they occur and the type of tree cover involved 
(Fig. 1.1). These major types of practice can be further classified in terms 
of the density and arrangement of the tree component (Fig. 1.2). This 
results in defining groups of practices which share important ecological 
and managerial characteristics. Lists and descriptions of common 
agroforestry practices can also be found in standard agroforestry texts 
(Nair, 1993; Young, 1997; Huxley, 1999). 

Trees and soil fertility 

Whatever the reasons farmers have for planting or protecting trees in a 
specific case, they nearly always fulfil several functions simultaneously. 
Trees may have been planted on a hillslope to produce timber or fruits, 
but they may also protect the soil from being eroded. Trees planted or 
retained for fodder are often nitrogen-fixing and may improve nitrogen 
availability in the soil. Similarly, trees that have been allowed to regenerate 
in a riparian zone because of environmental regulation, for firewood 
production or simply an appreciation of their scenic value may also filter 
nutrients from runoff water, thereby retaining them in the land-use system 
and protecting the river from eutrophication. 

This book is about the effects of trees on soil fertility. Soil fertility is 
defined here as the ability of a soil to serve as a suitable substrate on which 
plants can grow and develop. Fertile soils facilitate root development, 
supply water, air and nutrients to plants, and do not have pest and disease 
burdens that result in catastrophic impacts on the plants that are being 
grown. Maintaining soil fertility is the basis of all forms of sustainable land 
use, that is, land use that remains productive in the long term. If fertility 
has fallen below a critical level through long-term agricultural use without 
replacement of nutrients or as a result of erosion, or if it is naturally very 
low, the replenishment of soil fertility may be a precondition for productive 

Tropical soils are not generally infertile, but infertile soils are very 
common in the tropics. Moreover, recent research has provided evidence 
that soil fertility is decreasing in many farmed areas in the tropics. Nutrient 
budgets which were established at different spatial scales in sub-Saharan 
Africa, Ecuador and small farms in Costa Rica showed net nutrient losses 
from agricultural soils because of off-take in crop yields, leaching, erosion, 

npacts of Trees on Fertility 

runoff and gaseous losses, which were not matched by nutrient inputs from 
mineral and organic fertilizers, atmospheric deposition, biological nitrogen 
fixation and, in flooded or irrigated areas, sedimentation (de Koning et 
al., 1997; Smaling et al., 1997; Stoorvogel and Smaling, 1998). Negative 
nutrient budgets like these are especially threatening for the fertility of 
soils whose nutrient stocks are already small, such as sandy soils with low 
organic matter contents, which are widespread in African savannas, or acid 
soils, which occupy vast areas of the humid tropics of Latin America, Africa 
and Asia (von Uexkiill and Mutert, 1995). Accordingly, long-term fertility 
studies on farmers' fields in African savannas have revealed evidence of 
widespread chemical and physical soil degradation, including negative soil 
organic matter and nutrient balances, although these have not always 
immediately been translated into declining crop yields (Pieri, 1989). Low 
and declining soil fertility are recognized by many tropical farmers as 
major constraints to agricultural production (Smaling et al., 1997). It can 
be expected that projected growth of human populations in tropical 
countries will further aggravate these problems, especially when 
population pressure increasingly obliges farmers to cultivate fragile and 
naturally infertile soils which are particularly prone to degradation. 

Both scientific research and farmers' observations clearly point to the 
need to improve current farming practices with respect to their ability to 
increase and sustain soil fertility and agricultural productivity. The green 
revolution attempted to increase agricultural productivity in the tropics 
through increased inputs of mineral fertilizers, pesticides and new crop 
varieties. Although the successes were sometimes spectacular on relatively 
fertile land with good infrastructure, large numbers of small farmers in 
marginal environments were bypassed by these developments because they 
could not afford the necessary investment in new seeds and chemical 
inputs. Gradually, it has become understood that more accessible means 
are needed to enable many small farmers to feed their families and raise 
their living standards, and it has been suggested that this is most likely to 
come about from a thorough understanding of the biological bases of soil 
fertility (Woomer and Swift, 1994). Trees with their numerous beneficial 
effects on soil fertility play an important role in this strategy. 

The fundamental assumption in agroforestry, that the integration of 
trees into farming systems and landscapes can increase soil fertility, 
productivity and sustainability, was initially based mainly on the 
observation that soil under forest vegetation generally remains fertile and 
that tree fallows are able to regenerate degraded soils, as occurs in shifting 
cultivation systems (Nair, 1984). Subsequent scientific research has 
increasingly produced insights into the mechanisms through which trees 
improve soil fertility, though many key processes have still not been fully 
quantified. Agroforestry practices have been shown to influence chemical, 
physical and biological components of soil fertility. Trees can improve the 

C. Schroth and F.L. Sinclair 

nutrient balance of a site both by reducing unproductive nutrient losses 
from erosion and leaching and by increasing nutrient inputs through 
nitrogen fixation; they can improve soil structure, water-holding capacity 
and crop rooting volume; and they can increase the biological activity in 
the soil by providing biomass and a suitable microclimate. However, a 
better understanding of the interactions between trees and soils has also 
helped to keep expectations at a realistic level and to recognize what 
agroforestry can and cannot achieve. Many commonly measured nutrient 
fluxes in agroforestry systems are part of the internal nutrient cycling 
within a system and do not change its overall nutrient budget. If a site is 
deficient in nitrogen and phosphorus, leguminous trees may be able to 
increase the availability of nitrogen through biological nitrogen fixation, 
but phosphorus may have to be added from external sources. 

It has also become increasingly clear that the intensity with which trees 
and tree-crop associations influence soil fertility differs widely between 
agroforestry practices, even if the processes are similar in principle. For 
example, the nitrogen fixation and biomass production of relatively few 
leguminous trees dispersed in a crop field will be much less than those of 
a closed stand of these trees in a planted fallow. Also, most trees will take 
up some of their nutrients from the subsoil and deposit them in surface 
soil through leaf litter and root decay and thus act as a nutrient pump. 
However, for some combinations of species managed in particular ways, 
this process may be important, whereas for others the effects may be too 
small to be of much consequence. When designing or improving 
agroforestry techniques, it is therefore important that the technique is 
matched with the fertility problems that are seen as priorities at a given 
site, rather than assuming that every type of agroforestry will improve soil 
fertility in general. 

Matching an agroforestry technique to the biophysical aspects of a site 
is necessary but not sufficient to ensure adoption; it also has to be 
compatible with the views, experiences, traditions and economic capacities 
of the farmers. Beneficial effects of trees on soil fertility are often only 
perceptible after several years and small farmers often cannot afford to 
invest in tree planting and tending without receiving an immediate return. 
Some techniques may require more time for pruning or biomass transport 
than the farmer can afford, especially when these activities are necessary 
at times when sowing, weeding or harvesting of crops are more urgent 
needs. It is also important to recognize that sustainable production from 
the same piece of land on the basis of stable soil fertility is not always a 
primary objective of the land user. Where the effort for clearing a new 
piece of land is smaller than that of maintaining the productivity of the 
already cleared plot, or where land clearing is a way of establishing 
ownership rights or is advantageous for other reasons, investments in 
sustainability may not have a high priority. Similarly, where runoff and 

npacts of Trees on Fertility 

erosion from hillslopes benefit valuable crops in the valley below, soil 
degradation on the slopes may be seen as an acceptable price to pay for 
high productivity in the valley. It is thus clear that progress in agroforestry 
depends on a thorough understanding of both the biophysical and 
socioeconomic dimensions of farming systems at a range of scales. 

1 .2 Objectives and Structure of the Book 

This book provides an overview of the principal concepts that have been 
developed over the past decades concerning the effects of trees on soil 
fertility and an in-depth discussion of the methodological approaches that 
are appropriate to their study. It has been written mainly for researchers 
and students interested in tropical agroforestry. Case studies, examples 
and references have been taken mostly from the tropical agroforestry and 
ecology literature. However, although temperate agroforestry differs in 
its socioeconomic context from that in the tropics, the general biophysical 
processes are similar, and so most of the information presented here is 
also relevant to temperate conditions. Moreover, trees and soils in 
agroforestry are not fundamentally different from trees and soils in forests 
and especially in savannas, and we expect that students and researchers 
in forestry, agronomy and ecology will also find much useful information 
in the following chapters. 

Included in the book are economic, chemical, physical and biological 
aspects of soil fertility. Because of the integrated presentation of theoretical 
concepts and research methodologies, the book is particularly suitable for 
people who intend to do practical research on the interaction between 
trees and soils. It will be most useful to those who already have some basic 
knowledge of both agroforestry and soil science, although there are 
comprehensive references to the literature in these areas from which such 
an understanding could be gained. 

The chapters begin with synopses that are followed by methods 
sections. The synopses provide concise statements of essential background 
knowledge and outlines of the principal hypotheses and research results 
relevant to the topic at hand, and suggest key areas for future research. 
These sections are intended to help researchers in the identification of 
suitable, worthwhile topics for their research, and for students to see 
individual research results within the context of the larger body of 
knowledge and hypotheses relating to tropical soil fertility and 
agroforestry. The methods sections not only include those methods that 
are currently used in agroforestry, but also methods whose utility has been 
demonstrated in other fields and that could and should be applied in 
agroforestry studies in the future. 

10 C. Schroth and F.L. Sinclair 

This is not a methods book in the conventional sense, as it does not 
provide detailed field and laboratory descriptions of research methods. It 
has been produced to augment rather than to replace established methods 
texts, such as Tropical Soil Biology and Fertility: a Handbook of Methods 
(Anderson and Ingram, 1993), by furnishing a broader discussion of the 
scientific background of soils research in tropical agroforestry and the 
range of available research methods. The book also provides extensive 
reference to the relevant methodological literature, both from agroforestry 
and from soil science in general. 

The book starts with a chapter on the economics of soil fertility 
management and agroforestry practices, in which concepts for the analysis 
of farmers' decisions are presented and consequences for the adoption or 
non-adoption of agroforestry practices are discussed. Attention is drawn 
to the substantial public benefits at national and global scales that derive 
from the ecosystem services provided by trees. At present farmers often 
bear the costs of these public goods, which results in a level of agroforestry 
adoption that may be lower than is desirable from a societal point of view 
(Chapter 2). Chapter 3 presents some general grounding in appropriate 
methods for experimentation, sampling and data analysis for soil fertility 
research in agroforestry. A section on fallow experimentation is included 
because of the particular requirements of working with rotational systems 
that have distinct temporal phases and their present importance in 
agroforestry research. There is also a section on geostatistics, a tool that 
has not yet been widely used in agroforestry research but may become 
increasingly important in the future because of its usefulness in the analysis 
of spatial heterogeneity. Chapter 4 discusses the dynamics of organic 
matter in tropical soils as influenced by agroforestry and other land-use 
practices. The following four chapters are dedicated to different aspects 
of nutrient cycling. The synopsis section of Chapter 5 discusses the 
principal hypotheses concerning how trees may affect nutrient cycles and 
introduces the concepts of competition, complementarity and facilitation. 
This discussion provides a framework within which the macronutrients 
nitrogen, phosphorus, potassium, calcium and magnesium as well as soil 
acidity are discussed. The next three chapters treat nutrient cycling 
processes of particular relevance to agroforestry: decomposition and 
nutrient release from biomass (Chapter 6), nutrient leaching (Chapter 7) 
and nutrient capture (Chapter 8). Atmospheric nutrient inputs and 
nutrient losses into the atmosphere through fire are the topics of Chapter 
9. Physical soil fertility is treated in the following two chapters on soil 
structure (Chapter 10) and soil water (Chapter 11). Five chapters are 
dedicated to biological aspects of soil fertility: root systems (Chapter 12), 
biological nitrogen fixation (Chapter 13), mycorrhizas (Chapter 14), 
rhizosphere processes (Chapter 15) and soil fauna (Chapter 16). Soil 
erosion is the topic of the last chapter (Chapter 17). Throughout the text 

Impacts of Trees on Fertility 1 1 

some effort has been made to demonstrate that the chemical, physical and 
biological components of soil fertility strongly interact. 

Topics related to soil fertility and agroforestry that have not been 
included in this book include salinization and waterlogging, which have 
been addressed in several contributions to a recent symposium (Lefroy et 
al., 1999); soil-related pest and disease problems, which are still a neglected 
field in agroforestry despite recent progress (Desaeger and Rao, 1999; 
Dupomiois et al, 1999); and allelopathy, on which a large literature exists 
(e.g. May and Ash, 1990; Ramamoorthy and Paliwal, 1993; Conger, 1999), 
although there remains a paucity of information about its practical 
importance in agroforestry. 

Chapter 2 

Economic Aspects of Soil Fertility 

Management and Agroforestry Practices 

(A.-M.N. Izac) 

World Agroforestry Centre, International Centre for Research in 
Agroforestry (ICRAF), PO Box 30677, Nairobi, Kenya 

2.1 Introduction 

Analyses of adoption of soil fertility management and agroforestry 
practices by farmers in tropical countries are relatively sparse. They 
provide a patchwork picture of cases where adoption occurred relatively 
rapidly and was quite widespread and other cases where adoption did not 
occur on any significant scale. This chapter deals with the economic 
concepts that can be used by scientists interested in increasing the 
probability of adoption of soil fertility management practices by farmers. 
It is stated in Chapter 1 of this book that: 'When designing or improving 
agroforestry techniques, it is ... important that the technique is matched 
with the fertility problems that are seen as priorities at a given site, rather 
than assuming that every type of agroforestry will improve soil fertility in 
general.' The goal in this chapter is to present economic concepts for 
matching soil fertility interventions to both farmers' constraints and 
objectives and those of society, in order to increase their likelihood of 
adoption. The chapter highlights key economic concepts and processes 
that biophysical scientists should find helpful in putting their own work 
into the broader context of the farmer's economic environment. It does 
not attempt to provide detailed discussions of economic approaches to the 
assessment of agroforestry practices for professional economists 
(economists are not the major intended audience for this book). 
Furthermore, such detailed discussions are not readily available in the 
literature, and therefore may constitute a gap which needs to be filled. 
Declining soil fertility is acknowledged as a problem by the vast 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 1 3 

F.L. Sinclair) 

14 A.-M.N.Izac 

majority of farmers who experience it on their farms (see Chapter 1). It 
is, at the same time, a problem for society as a whole as it is related to issues 
of agricultural sustainability, soil biodiversity, carbon sequestration and 
watershed functions. This chapter focuses on the decision-making 
processes that determine whether an agroforestry intervention will be 
adopted by farmers and whether ensuing levels of adoption will be socially 

These processes are analysed within the framework of ecological 
economics, which differs substantially from mainstream economics. The 
emerging field of ecological economics addresses the relations between 
ecosystems and social and economic systems. Social and economic systems 
are viewed as subsystems of the biosphere and thus as wholly dependent 
upon ecological-economic interactions. This is in direct contrast with the 
conventional or mainstream economic approach, which considers that all 
phenomena are subsumed within and obey the rules of an economic 

In what follows, factors affecting farmers' decision-making processes 
regarding adoption of soil fertility and agroforestry practices are first 
highlighted. A brief discussion of agricultural systems hierarchy serves to 
set the context within which farm-scale decision-making processes are 
analysed. Finally, economic processes relevant at the regional and global 
scales are discussed. 

2.2 Factors Influencing Farmers' Decisions About Soil 
Fertility Management Practices 

The decision to adopt a given soil management technology is made by an 
individual farmer on the basis of a number of factors which the farmer 
integrates into a framework driven by the farmer's production objectives. 
Social scientists who have analysed farmers' decision making in the tropics 
have shown that farmers think in a systemic fashion. Decisions regarding 
a given field or a given practice, such as soil fertility management and 
agroforestry, are thus not made in isolation. These decisions are made 
within the context of the whole farm and of the totality of the resources 
and assets available to the farmer. These resources and assets include: 
(i) labour (family labour plus hired labour if sufficient cash is available); 
(ii) cash to buy fertilizer and other chemicals; (iii) their entire landholding 
and the different fields comprising it; (iv) purchased assets such as 
implements, machinery, animal traction; (v) access to water (either on farm 
or off farm); and (vi) access to other off- farm resources (such as communal 
resources, forested lands and woodlots). 

Farmers focus on the trade-offs between the efforts they have to make 
to meet their production objectives (being able to produce enough food 

Economic Aspects of Soil Fertility Management 1 5 

for the family, being able to produce a surplus and to sell it) and the payoffs 
they expect from these efforts. They consider the totality of their holding 
and its various current and potential uses vis-a-vis their production 
objectives and weigh these in terms of the outcomes they expect when 
combining their resources into different practices (including soil fertility 

Farmers also have to factor into such decisions different groups of 
markets and different types of customs. They need to consider markets 
for agricultural commodities (local, regional, national and international if 
relevant), since the sale of their surplus production, over and above what 
they need to produce to feed their family, depends on these markets. They 
also have to take into account markets for inputs. These determine the 
costs they will have to incur for their use of: (i) labour, be it hired or family 
labour; (ii) land, especially if they rent some of their fields; (hi) capital, 
especially if they borrow money; (iv) implements and machinery; and (v) 
water and other resources. The level of income of farmers is determined 
by these two groups of markets. Finally, the actual purchasing power of 
this income is dictated by markets for consumer goods (such as clothing 
and medicines) and by government policies regarding public 
infrastructure (such as means of transport and their costs, costs of 
education, health and electricity). 

Social customs and norms influence a number of the elements that 
farmers need to integrate in their decisions. These customs and norms 
determine farmers' access to many natural resources, as well as human 
labour, through the prevailing land and tree tenure system, and the 
regulations concerning access to off-farm resources (water, communal 
resources such as forests and woodlots), and access to non-family labour. 
In some societies in western Africa for instance, rules of access to labour 
are extremely complex and based on what anthropologists call reciprocal 

It is important for scientists undertaking research on the development 
of soil fertility management practices to be cognizant of the fact that 
farmers' decisions to adopt such practices are complex and driven by the 
conjunction of the above factors. Scientists need to understand the basic 
principles at play in such decisions in order to design interventions that 
match farmers' constraints and are therefore adoptable. 

The various types of factors influencing farmers' decisions are depicted 
in Fig. 2.1. Figure 2.1 also illustrates the fact that these factors are 
determined at different spatial scales. Although some of them are clearly 
determined at the farm scale (e.g. family labour and soil fertility status of 
the fields) others are determined at the landscape and community scale 
(e.g. access to off-farm resources) or at the regional and national scale (e.g. 
government policies concerning rural infrastructure and markets for 
consumer goods). 


A. -M.N. Izac 

Social-political parameters 

• political system 

• social customs 

• land and resource tenure 

• demography 

• infrastructure (transport, 
credit, education, health) 

Economic parameters 

• markets for inputs and outputs 

• degree of integration in markets 

• non-farm income 

• technologies available 

• existence of a surplus from 
previous year 

• endowment in family labour, 
land, capital 

Ecological parameters 

• climate 

• soils 

• hydrology 

• pests 

• weeds 

• topography 

Farmers' decision making 

• what to produce 

• when to produce it 

• how to produce it (including soil fertility 
management practices) 

Outputs produced 

quantity and quality 
(crops, animals, trees...) 

Environmental impacts 

depending upon practices chosen: 

soil erosion, acidification, water pollution, 

deforestation, decreased genetic diversity, or. 

enhanced natural resource base, sustainable 


Fig. 2.1 . Context of farmers' decisions to adopt. 

Finally, Fig. 2.1 shows that a decision to use specific practices, as per 
these factors and per the farmer's production objectives, have specific 
outcomes for the farming household and may also have off-farm 
consequences. The off-farm consequences include, for instance, the forest 
degradation that would result from farmers deciding to ameliorate the 
effects of decreasing soil fertility on their fields by transferring forest soil 
to their land, as occurs in parts of India. Another example would be the 
increased water sedimentation for downstream farmers resulting from a 
decision to clear secondary fallow land for agriculture by upstream farmers 
in northern Thailand. 

The purpose of this section is to debunk some often-held myths about 
farmers' decisions concerning soil management in tropical countries. The 
following points are stressed. First, farmers in tropical countries are as 
rational in their decision making as any other persons or stakeholders in 
any country. Secondly, they generally have a different perspective from 

Economic Aspects of Soil Fertility Management 


that of scientists when considering improved soil management practices. 
Theirs is a system perspective whereas scientists often adopt a reductionist 
perspective when designing technologies. This may explain why rates of 
adoption of these technologies by farmers are often a disappointment to 
scientists who have not factored into their technology design all the 
relevant constraints faced by the farmers. Thirdly, farmers' decisions to 


Climate, geology, etc. 

Large geographical 
and economic units 

National economic 
policies and 

International trade 

^..- — ' 

~ — — — ___ 



Natural resources 
endowments, human 
resources endowments 

Rural regions 

Industrial and 
urban regions 

Regional institutions, 


~ — — — __. 


Topography, pest 
population, etc. 

Farming households 


Social norms and 
customs, roads, 



" — - — - , . 


Various inputs 


making units 



Crop production 

,_— . — • — 

■™— — — . 














Fig. 2.2. Hierarchy of systems within which farmers make decisions. 

A. -M.N. Izac 

adopt are constrained by various factors, determined at different spatial 
scales; likewise these decisions to adopt can have consequences at spatial 
scales beyond the farm scale. 

2.3 Hierarchy of Agricultural Systems as a Background to the 
Understanding of Farmers' Constraints 

Figure 2.2 represents a spatial hierarchy of agricultural systems within 
which farmers' decisions are made. The highest level in the hierarchy, 
supraregional systems, occupies the largest land area and can transcend 
national boundaries. Macroeconomic processes, as well as certain 
geological processes, are best understood at this level. The lowest level (soil 
systems) covers the smallest spatial unit and is the level at which specific 
biological processes such as nutrient uptake may be investigated. A basic 
rule in systems theory is that systems at level n are constrained and 
controlled by systems at level n + 1 , and in turn they constrain systems at 
level n - 1 (Allen and Starr, 1982). 

Farmers (operating at the farming system level) thus have to take the 
environment provided by the village or landscape level as a constraint in 
their decisions regarding soil fertility practices. Likewise, the village or 
landscape scale is constrained by regional system variables that operate 
themselves within the confines of supraregional variables. Consequently, 
farmers integrate a wide range of ecological, social and economic 
parameters belonging to levels higher than the farming system in their 
decisions to adopt soil fertility management and agroforestry practices. 

Decisions made at the farming system scale have repercussions at the 
same scale, as well as at lower and higher scales in the hierarchy. These 
are mediated through various economic and biological processes, such as 
nutrient cycling and the market mechanism. Because these processes 
transcend farm boundaries, it is helpful to establish a distinction between 
the economic processes that occur at the farm scale and those that are 
manifest at the landscape or watershed and global scales. Even though soil 
fertility management and agroforestry practices are very localized 
interventions on farmers' fields, it is essential for scientists to realize that 
the key processes of relevance to their adoption occur at the farm, regional 
and global scales. 

2.4 Anatomy of a Decision at the Farm Scale and Economic 
Methods for Understanding such Decisions 

The decision by an individual farmer to adopt a given soil fertility 
management or agroforestry practice is made within the frame of the 

Economic Aspects of Soil Fertility Management 1 9 

factors shown in Fig. 2.1. More specifically, such decisions are based on 
farmers' production objectives and farmers' perceptions of the advantages 
and disadvantages of a given practice. These advantages and disadvantages 
are all the monetary and non-monetary costs and benefits of a practice, as 
perceived by farmers. 

All these costs and benefits are relative to, or determined by, the 
existing land tenure system and markets. For instance, Unruh (2001) has 
shown that in post-war Mozambique, where land rights are unclear and 
ambiguous, land disputes are very common and costly. Agroforestry trees, 
in particular older cashew trees (Anacardium occidentale), are, however, 
considered as evidence of land ownership. In such an institutional context, 
these trees have a substantial non-monetary benefit; they serve to clearly 
establish stronger land claims for farmers (Unruh, 2001). Notions of costs 
and benefits are thus highly relative concepts in economics. What is a 
benefit in a given institutional context, such as trees that establish land 
rights in Mozambique, may not be so in countries where land rights are 

In what follows, the basic economic principles and processes at play 
in farmers' decisions to adopt or not to adopt are analysed, and examples 
from actual economic analyses are given to illustrate the argument. The 
'real-world' costs and benefits of a given agroforestry practice in a given 
location are highly site specific and variable across farms. This is a direct 
consequence of the complex range of factors that determine the value of 
each cost and benefit. It is thus very important for scientists interested in 
understanding farmers' decisions to be aware of the economic processes 
that guide these decisions. Without such an understanding, empirical 
evidence of the profitability (or lack thereof) of specific agroforestry 
options may appear very confusing. 

Principles of private cost-benefit analysis and net present value 

One of the simplest ways to assess these costs and benefits is to use private 
cost-benefit analysis. This analysis consists of comparing the flows of 
benefits and costs generated over time by the adoption of a given 
agroforestry practice. Only the costs and benefits that are relevant to a 
given farmer are taken into consideration, and they are valued at the 
market prices faced by the farmer. No attempt is made to adjust costs and 
benefits for market and government failures. Nor is any attempt made to 
take social costs and benefits (externalities and environmental benefits) 
into consideration. Pagiola (1994) and Sugden and Williams (1978) 
provide excellent introductions to cost-benefit analysis for non-economists. 
Gittinger (1982) is one of the classic texts on cost-benefit analysis, along 
with Mishan (1976). Both provide much detail about the approach, its basic 

20 A.-M.N.Izac 

assumptions, and various methods of computing costs and benefits. Both 
texts are written for professional economists. 

The basic principle in private cost-benefit analysis is simple. The net 
value, in today's dollars, of the future flows of benefits and costs of an 
agroforestry option is computed. In economic language, this is called the 
net present value of an agroforestry option. This net present value (NPV) is 
defined as: 

B t ~C t 

NPV = X^7 (2.1) 

(l + r) 

where: B t = benefits in year t; C t = costs in year t; r = discount rate, that 
is, the rate at which benefits and costs incurred in year t have to be 
discounted to arrive at their value in today's dollars; and n = number of 
years during which the agroforestry option will generate benefits and costs. 
In what follows, the concepts of costs and benefits are further 
discussed, as well as the concept of discount rate. 

Monetary and opportunity costs of adoption 

The generic soil fertility and agroforestry practices that are discussed in 
the other chapters of this book include improved fallows (tree-crop 
rotations), tree-crop associations (such as hedgerow intercropping and 
shaded perennial crops), contour hedgerows and boundary plantings. 
Each of these options entails various monetary and non-monetary costs 
for farmers. Therefore: 

C t = mc t + oc t (2-2) 


here mc t = sum of monetary costs in year t, and oc t = sum of opportunity 

costs in year t. 

The monetary or out-of-pocket costs (mc t ) are quite straightforward. 
They consist of: (i) the costs of buying seeds and/or seedlings (or the costs 
of having an on-farm nursery for the tree seedlings), and whatever other 
inputs such as fertilizer are used, and (ii) the costs of hiring additional 
labour (if family labour is insufficient) for planting the trees, weeding the 
trees, pruning them, and applying or incorporating tree primings to the 
soil (see Table 2.1). The most meaningful way of assessing these costs is on 
the basis of on-farm experiments or under actual farming conditions, 
through on-farm monitoring of farmers' practices. Monitoring the use of 
hired labour throughout the year, and throughout the lifespan of an 
agroforestry option is not an easy task, and is costly when this lifespan is 
long. There are therefore relatively few examples of private cost-benefit 

Economic Aspects of Soil Fertility Management 21 

Table 2.1. Generic on-farm costs of agroforestry practices. 

• Monetary costs of additional hired labour for planting tree seedlings/seeds, 
pruning the trees, weeding the trees (most agroforestry options) 

• Monetary cost of purchase and transport of seeds and seedlings and fertilizer 
(if applied; most agroforestry options) 

• Opportunity cost of not using a field or part of a field in another way (for 
improved fallows, intercropping, alley cropping and trees on contour ridges) 

• Opportunity cost of family labour used (forgone participation in other farming 
activities; most agroforestry options) 

• Opportunity cost of possible losses in germination of crops (for agroforestry 
options requiring the application of tree residues to crops) 

• Opportunity cost of lower crop yields due to competition from trees (for 
intercropping, alley cropping and parklands) 

analyses that use such 'real-world' data. Many analyses rely on on-station 
experiments and on extrapolations of on-station data to on-farm 
conditions. Dewees (1995) provides an interesting analysis of the costs to 
farmers in Malawi of intercropping maize with Faidherbia albida, and of 
alley cropping Leucaena leucocephala with maize. 

The non-monetary or opportunity costs (oc t in Eq. 2.2) are not actual 
disbursements for farmers, but are nevertheless very real costs which 
farmers take into consideration in their decisions. They consist of all the 
opportunities for generating benefits that a farmer gives up when choosing 
a given agroforestry option (see Table 2.1). The appropriate baseline for 
such comparison is the best alternative practice that the farmer could have 
chosen in lieu of this agroforestry option. If there is an alternative practice 
that would bring higher benefits to the farmer than agroforestry, then it 
is highly unlikely that the farmer will choose agroforestry, unless 
agroforestry also brings about significant intangible social benefits (such 
as increased social status in the community) for the farmer. 

Two principal types of opportunity costs of agroforestry options are 
those of labour and land. When family labour is used for planting 
hedgerows, for example, the opportunity cost is equal to the benefit that 
this family labour could have generated if instead of planting seedlings it 
had been engaged in the best alternative, such as planting or weeding food 
or cash crops. The opportunity cost of family labour can sometimes be a 
major constraint to the adoption of agroforestry practices because: 

• the extra labour required for implementing some of these practices 
may be needed at a time of year when all household members are 
occupied in tasks given a high priority (e.g. production of food crops); 

• available family labour is decreasing in some areas due to HIV/AIDS; 

22 A.-M.N.Izac 

• in some countries such as Kenya there is a fairly strict division of on- 
farm responsibilities between females and males, so that even if female 
labour is plentiful some tasks such as pruning trees in hedgerows may 
not be done because male labour is insufficient (see Swinkels and 
Franzel, 1997, for details). 

In such cases, private cost-benefit analysis in itself may not be sufficient 
in capturing these non-quantifiable constraints, and farmer surveys 
will be needed to complement the picture painted by private cost- 
benefit analysis. It could indeed be the case that a given agroforestry 
option, although economically profitable, is at the same time not really 
feasible, and therefore will have a low adoption rate, for the reasons just 

Whereas monetary costs of adoption are simply assessed by their 
market value (e.g. cost of one seedling times number of seedlings 
purchased; hourly wages paid to hired labour times number of hours 
worked), opportunity costs are more difficult to assess. In the case of family 
labour, economic theory indicates that the prevailing wage rate times the 
number of hours worked by family labour should be used to assess this 
cost. This is because it is assumed that, if family labour had not been 
working on the farm, it could have been engaged working as hired labour 
on other farms. When rural unemployment is high, however, it is doubtful 
that it would have been feasible for family members to be thus employed 
in the agricultural sector. As a rule of thumb, the higher the rate of 
unemployment in an area, and the more difficult for family members to 
travel within this area (because of geographical isolation or bad transport 
networks), the lower is the wage rate that should be used to assess the 
opportunity cost of family labour (see Pagiola, 1994, for a clear discussion 
of these issues). 

The opportunity cost of an agroforestry option in terms of the land 
area it occupies (entire field or portion of fields) is easier to assess. It is 
equal to the benefits the farmer would have gained from the best 
alternative land use. This is generally the dominant or traditional land use 
in an area, such as the cultivation of maize with a very low level of fertilizer 
in the case of improved fallows in western Kenya. In such an example, the 
opportunity cost of the land under improved fallows is equal to the net 
benefits the farming household would have received from the traditional 
maize cultivation system on a per hectare basis, times the land area under 
fallow. In some instances, the best alternative to the agroforestry option 
under analysis may be a different soil fertility enhancement method that 
farmers could choose to use (e.g. green manuring with legumes). In this 
event, net benefits from manuring with legumes on a per hectare basis 
times the land area under fallow will be equal to the opportunity cost of a 
fallow practice. 

Economic Aspects of Soil Fertility Management 23 

Not only are these various costs (represented in Table 2.1 and equal 
to C t in Eq. 2.2) entirely borne by the farming household, but also the 
household must start paying for them from the moment a practice is 
chosen for implementation. In addition, some of these costs will recur over 
a number of years. In some of the few assessments of actual costs of 
adoption incurred by farmers published in the literature, total costs of 
adoption decrease very substantially between year one and the following 
years, and are, during the first years, significantly higher than the costs of 
implementing farmers' current practices. This was the case, for instance, 
for hedgerow intercropping (maize with Leucaena leucocephala or Calliandra 
calothyrsus) in the highlands of western Kenya. The total costs of adoption 
to farmers (C t ) amounted, on average, to US$422 ha -1 during the first year 
and decreased, on average, to US$276 ha -1 during the second year 
(Swinkels and Franzel, 1997). This compared with total costs of US$296 
for the traditional maize system during the first year and costs of US$27 1 
during the second year. In another example, of improved fallows in Costa 
Rica (with Vochysia ferruginea and Hyeronima alchorneoides), actual costs (for 
clearing, planting, replanting, weeding, pruning, and seedlings and their 
transport) were around US$1330 ha" 1 for the first year. They decreased 
to US$162 ha" 1 during the second year and US$97 ha" 1 during the fifth 
year of adoption (Montagnini and Mendelsohn, 1997). 

Monetary and non-monetary (environmental and social) benefits of 

The benefits of agroforestry practices, as they are perceived by the farmer, 
are generally more difficult to assess than their costs. And, furthermore, 
some of these benefits often start occurring after a while, so that there is 
a time lag between the moment when a household incurs costs in order to 
adopt an agroforestry practice and the moment when it starts receiving 
benefits from this practice. 

In parallel with costs, the benefits of adoption of a practice can be 
monetary and non-monetary (see Table 2.2). Therefore: 

B t = mb t + ib t (2.3) 

where mb t = monetary benefits in year t; and ib t = non-monetary benefits 
in year t. 

There are two kinds of monetary benefits (mb t ). The first is the value 
of the increased crop yields associated with a given agroforestry practice. 
In a maize-based improved fallow system for instance, this will be the 
market value of the increased maize yields following the improved fallow 
period. Such direct monetary benefits will occur over a period of time, that 

24 A.-M.N.Izac 

Table 2.2. Generic on-farm benefits of agroforestry practices. 

Time 3 








Benefits Year 1 Years 2-5 Years 6-10 


Increased crop yields + ++ ++ 

Value of various tree products + ++ 


Increased resilience and sustainability of system through: 
Decreased risks of yield fluctuations with 

usual climatic variability 
Enhanced soil resource base 
Enhanced capacity of system to adjust to 
exogenous changes without 
generating increased flow of pollutants 
Increased biodiversity of soil biota 
Increased biodiversity of fauna and flora 

a The time scales shown are indicative. 

0, no measurable benefit; +, ++, benefit is measurable and its intensity varies 

from low (+) to high (++). 

is, until all residual effects of the improved fallow have been exhausted. 
To continue with our example, it is thus necessary to assess maize yields 
over time in the improved fallow system (as well as maize yields over time 
in the traditional maize system, since their market value will constitute the 
opportunity cost of the land under fallow). Nominal farm-gate prices of 
maize over the relevant period of time also need to be obtained or 
predicted. Even if the maize is grown for home consumption rather than 
for sale, the value of this monetary benefit of adoption will be evaluated 
as: tonnes of maize produced multiplied by the market (farm-gate) value 
of a tonne of maize, divided by the relevant number of years. 

The second type of monetary benefit is the market value of the diverse 
products that may be generated by an agroforestry practice. Examples of 
these products include indigenous fruits, timber, nuts, leaves that are used 
in food preparation, leaves, bark and seeds that have medicinal properties, 
poles, fodder, fuelwood. Proceeds from the sale of these products will 
contribute to household income and are thus a monetary benefit. If these 
products are not sold but are consumed by the household, their market 
value equivalent still represents a monetary benefit, since the household 
does not need to spend some of its income on the purchase of these 
products. In a very real way, these products directly contribute to the 
welfare of the household. Leakey and Tomich (1999) provide various 
examples of the market values of tree products from agroforestry systems, 

Economic Aspects of Soil Fertility Management 25 

although these systems are not restricted to those that enhance soil fertility. 
For all such monetary benefits, the effects of an agroforestry option on the 
yields of various products, from crops to medicinal products, need to be 
quantified on a yearly basis over the relevant time period. The value of 
these yields is then obtained by assessing or predicting the farm-gate prices 
of these products. 

To use again the example of hedgerow intercropping in western 
Kenya, total monetary benefits increased from US$709 ha" 1 during the 
first year to US$766 ha" 1 during the second year (Swinkels and Franzel, 
1997). This compared with total monetary benefits of US$707 ha" 1 for 
traditional maize; these benefits remained at this level over time. 

The non-monetary benefits of agroforestry practices at the farm scale 
(ib t in Eq. 2.3) are particularly difficult to assess. They consist of all the 
improvements in soil and other ecological processes that are brought about 
by the agroforestry practices and are not captured by increased crop yields. 
In other words, to understand and assess these non-monetary benefits, 
one needs to first consider all the functions which enhanced soil fertility 
(such as increased nutrient stocks, enhanced efficiency of nutrient cycling) 
will have on a farm and which are over and above those already translated 
into increased crop yields. 

One such function that farmers, including resource-poor households, 
particularly value is the increased resilience and sustainability, and 
therefore the increased risk-buffering capacity, of their systems. Recent 
poverty surveys undertaken by the World Bank (Kanbur and Squire, 2000) 
show that the poorer a farming household, the more important it is to this 
household to have effective strategies for coping with risk (environmental, 
climatic, economic). Ability to manage risks and adapt to change is actually 
something which poor farmers rank as a priority concern, on a par with 
increasing their income (World Bank, 1994; Kanbur and Squire, 2000). 
The next section shows that one method of evaluating this increased risk 
buffering capacity of agroforestry options is to use a lower rate of discount 
(r in Eq. 2.1). 

There are many other examples of non-monetized benefits of 
agroforestry practices, such as increased soil biodiversity, decreased 
erosion, increased water infiltration and recharge of underground water 
table, improved soil structure, enhanced capacity of systems to adjust to 
change without generating increased flows of pollutants, enhanced carbon 
sequestration. Although some of these will eventually result in increased 
yields (e.g. improved soil structure), others will not have any measurable 
effects on crop yields (e.g. improved recharge of the underground water 

Biological scientists have not reached a consensus about how to assess 
these various functions of enhanced soil fertility. It would thus be 
unrealistic to assume that farmers will be aware of all these functions. 

26 A.-M.N.Izac 

Furthermore, even if they were aware of them, it is unlikely that they 
would factor all of them into their decisions to adopt. This is because some 
of these functions, such as increased soil biodiversity, are likely to be of no 
special interest to them. 

Finally, non-monetary on-farm benefits of adoption can include 
aesthetic value, habitat for welcome wildlife and shade. Farmer surveys 
have shown that such benefits can be valued very highly by farmers 
(Scherr, 1995). 

Table 2.2 presents generic categories of benefits of agroforestry 
systems for soil fertility improvements (B t in Eq. 2.3) and illustrates a 
significant difference between the costs and benefits of adoption of such 
practices. This difference has been observed in a number of assessments 
of the costs and benefits of agroforestry practices (Current et al., 1995; 
Dewees, 1995; Montagnini and Mendelsohn, 1997; Swinkels and Franzel, 
1997). Most benefits are cumulative over time and reach greater amplitude 
3 or 5 years following adoption whereas costs have to be borne up front, 
as explained previously. 

Farmers' time horizons, discounting and balancing of the costs and 
benefits of adoption 

It was seen above that there are some fundamental differences between 
the costs and the benefits of agroforestry systems for soil fertility 
management at the farm scale. The costs are entirely borne by farmers 
and occur from the inception of adoption of a practice. The benefits are 
generally cumulative over time and low during the first years following 
adoption. In addition, farmers are unlikely to value some of the non- 
monetized benefits such as increased soil biodiversity. This raises the issue 
of the time frame of relevance to farmers. 

Recall that the NPV of an agroforestry option is represented in 
Eq. 2.1. This equation reflects the fact that benefits (and costs) occurring 
at some future time do not have the same value today as they will have 
in the future. Indeed, if we are asked whether we prefer to receive 
US$100 today or in 3 years' time, most of us will answer that we prefer to 
receive this sum today. This is because we can either spend the money 
now and immediately receive some enjoyment from it, or invest it to 
receive a greater amount in the future (US$ 100 plus interest over 3 years). 
The economic concept of a rate of discount (r in Eq. 2.1) is the measure 
which enables economists to translate future benefits and costs into today's 
monetary worth. It is the rate at which we need to be compensated in 
the future for willingly accepting not enjoying a benefit today. It is also 
called the rate of time preferences (see Mishan, 1976, for a detailed 

Economic Aspects of Soil Fertility Management 27 

As can be appreciated by simply looking at Eq. 2.1, the choice of a 
specific value for r is extremely important. Indeed, the value of NPV is 
highly sensitive to the value of r. When costs are incurred in the present 
and medium term, whereas benefits occur over the medium and long 
term, the lower the rate of discount used in the analysis, the higher the 
resulting NPV, and vice versa. 

Eliciting farmers' true rate of time preferences is a difficult task. 
Economic, psychological and social factors influence this rate. To arrive at 
a realistic rate through very carefully designed and in-depth interviews is 
a research project in and of itself. Consequently, in practice analysts use a 
rate, chosen on somewhat arbitrary grounds, that they assume reflects as 
well as possible the perspective of farmers in a given community. 

Resource-poor small-scale farmers in the tropics are highly vulnerable 
to risks (Kanbur and Squire, 2000) and generally concerned about their 
survival in farming from one year to the next. It is therefore likely that, 
even if they were aware of the medium- and long-term benefits of 
adoption, most farmers would value these future benefits relatively less 
than immediate benefits. This is because they occur over a period of time 
of little relevance to their immediate needs. In other words, the discount 
rate of relevance to farmers in tropical countries to be used in Eq. 2.1 is 
likely to be quite high (rates of around 20% are used in most studies). One 
empirical study of farmers' discount rates suggests that Costa Rican 
farmers discount the future at a rate of 20-25% (Cuesta et al., 1994). 
Another study of smallholders in the Philippines indicates that 40% would 
be an appropriate rate of discount for these farmers (Nelson et al, 1998). 
It should be noted that, when an agroforestry option is seen by farmers as 
a method for managing risk (such as farmers in Rajasthan who cope with 
repeated droughts by the complementary use of woody perennials and 
annuals), then their implicit rate of discount decreases substantially 
(Arnold, 1997), and rates of 5-10% can be used. 

Given the difficulties associated with the choice of an exact rate of 
discount and the sensitivity of the resulting NPVs to this choice, a good 
analytical practice is to use a range of rates. The corresponding range of 
NPVs will highlight the threshold rate of discount at which an agroforestry 
option becomes privately profitable for farmers. 

The relatively few empirical assessments of the NPV of different 
agroforestry practices generally indicate a positive NPV. For instance, in 
the 21 agroforestry projects implemented in eight countries of Central 
America and the Caribbean that were reviewed by Current et al. (1995), 
NPVs of trees with crops, alley cropping, contour planting, perennial crops 
with trees, homegardens and taungya systems were all positive, whereas 
the NPV of woodlots was negative at a 20% discount rate, indicating net 
economic losses for farmers. In another study on the loess plateau in 
China, the NPV of an agroforestry intervention consisting of apple 

A. -M.N. Izac 

orchards on bench terraces was positive, and was also about 100% higher 
than the NPV of soil conservation practices without a tree component (Lu 
and Stocking, 2000). 

Comparing the NPV of an agroforestry option to that of alternative 
options for farmers is an important step in understanding decision making 
at the farm scale. Various studies suggest that agroforestry systems typically 
have higher NPV than current or traditional practices (e.g. Current et al., 
1995, for Latin America and the Caribbean and the above example from 
China), with some notable exceptions in Africa (e.g. Dewees, 1995, for 
intercropping and alley cropping in Malawi; Drechsel et al., 1996, for 
improved fallows in Rwanda; Swinkels and Franzel, 1997, for alley 
cropping in Kenya). 

In addition to ensuring that the over time private profitability of an 
agroforestry option is compared to other options of relevance to farmers 
interests, it is also important to assess the sensitivity of this measure of 
profitability to changes in some variables. Sensitivity analysis consists of 
varying different parameters in Eq. 2.1, such as the price of labour, the 
yields and prices of the crops produced, to assess the effects of these 
changes on NPV (see Gittinger, 1982, for further details). The data used 
to assess these parameters have a built-in uncertainty, given the paucity of 
relevant secondary databases in tropical countries, the unreliability of 
existing data and the fact that primary data are costly and often difficult 
to collect (see Lele, 1991, for a discussion of these issues). It is therefore 
important to investigate the effects of changing the values of these 
parameters, within reasonable bounds, on the results of the analysis to 
better appreciate the robustness of the profitability of an agroforestry 
option. The study by Current et al. (1995) indicates that the positive and 
high NPVs obtained for the vast majority of agroforestry practices in 
various Latin American and Caribbean countries were most sensitive to 
changes in yields and product prices. In the African context, it would 
appear that results of NPV are most sensitive to changes in adoption costs 
and yields (e.g. Dewees, 1995; Drechsel et al., 1996). 

The absolute value of the NPV for an agroforestry option thus 
provides much information about the private profitability of the option, 
over its life span. The quality of the information is increased when a range 
of discount rates is used, when NPVs between different options are 
compared and when sensitivity analysis is undertaken. The concept must 
be further interpreted and analysed in order to also provide information 
about the adoptability of an agroforestry option, in addition to its private 

Economic Aspects of Soil Fertility Management 


Interpretation of results from private cost-benefit analyses 

Since NPV is an aggregate estimate of profitability over time, it is essential 
to identify the length of time during which net losses will be incurred by 
adopters before NPV starts becoming positive. This is called the break-even 
point of an agroforestry option. The evidence reported in the literature 
indicates that many agroforestry practices have break-even points longer 
than 1 year. An example concerns the adoption of trees on contour lines 
in Java. It takes a minimum of 5 years for the cumulative on-farm benefits 
to exceed the initial cost outlays in these agroecosystems (Barbier, 1990). 
In the example of adoption of improved fallows in western Kenya, it takes 
4 years for this to happen (Swinkels and Franzel, 1997). In their review of 
multiple agroforestry practices in several countries, Current et al. (1995) 
found break-even points ranging from just under 2 years (for alley 
cropping) to more than 9 years (for woodlots). Adoption of agroforestry 

total costs of 

TC, = TB 

Year 1 

Year 3 

\— ► 

Year 1 

Fig. 2.3. Farm-scale costs and benefits of agroforestry adoption over different time 

30 A.-M.N.Izac 

options that entail losses during 2 to 5 years (these are the most common 
break-even points reported in the literature) thus requires that farmers 
have the financial ability to absorb these losses. Clearly, not all resource- 
poor farmers in tropical countries will have such an ability. 

The soil fertility management and agroforestry practices that have the 
highest likelihood of being adopted by these farmers are those that result 
in sufficiently significant short-term benefits to compensate for the costs 
of implementation. This is summarized in Fig. 2.3, in which these costs of 
implementation are shown as a decreasing function over time. The 
corresponding benefits curve however, increases exponentially with time, 
for the reasons mentioned above. 

Two scenarios are represented in Fig. 2.3. In scenario a, total benefits 
(TB a ) increase rapidly over time whereas in scenario b they increase 
relatively slowly. If the planning horizon of the farmer is up to 3 years and 
if scenario a prevails, then adoption of agroforestry practices can be 
expected. If scenario b prevails and the farmers' planning horizon is 
around 3 years, then adoption is unfeasible because, over this time horizon, 
total benefits (TB b3 ) remain below total costs. This balancing of monetary 
and non-monetary, present and future, on-farm costs and benefits over 
the duration of the farmer's planning horizon is a fundamental economic 
process affecting soil fertility management at the farming system scale. The 
few economic analyses of agroforestry practices for soil fertility manage- 
ment that rely on actual on-farm data provide an illustration of this when 
they stress that a constraint to adoption is that farmers often do not have 
sufficient capital to be able to withstand the few years of low, or even 
negative, profits which precede the years of relatively high profitability 
(e.g. Scherr, 1995; Grist et al., 1999). 

It is enlightening to note that even in the economic context of the USA, 
where farmers receive government subsidies (called cost-sharing), tax 
credits and deductions for adopting agroforestry practices, rates of 
adoption of agroforestry practices have been lower than expected for 
essentially the same reasons. A survey of farmers in the southern USA 
revealed that, although these farmers were aware of the environmental 
benefits of agroforestry, 'considerable uncertainty remained' about the 
potential profitability of various agroforestry systems in the minds of these 
farmers (Zinkhan and Mercer, 1997). 

In summary, the fundamental economic processes that occur at the 
farm scale are as follows. 

• Farmers bear all the costs of adoption of improved agroforestry and 
soil fertility management practices, and these costs start occurring from 
inception of adoption and continue over time. 

• Farmers accrue the monetary benefits of adoption and some of its non- 
monetary benefits. 

Economic Aspects of Soil Fertility Management 31 

• These benefits are cumulative over time and tend to be relatively low 
during the initial years following adoption. 

• The time lag between costs and benefits of adoption for farmers, 
coupled with the fact that small-scale resource-poor farmers have 
short-term planning horizons and high discount rates, implies that rates 
of adoption can be very disappointing to scientists and policy makers. 

• A major implication for scientists is that it is essential for them to design 
practices which minimize up-front costs of adoption (labour, capital 
and land). 

The simplest and most effective economic method available for 
quantifying these processes for specific agroforestry practices is private 
cost-benefit analysis. It encompasses the computation of the NPV of a 
practice, from the perspective of an individual farmer. Very useful insights 
into the probability of adoption of a practice can be gained by: 
(i) comparing the NPV of an agroforestry option with the NPV of other 
land-use options of relevance to farmers; (ii) undertaking a sensitivity 
analysis of NPV; and (iii) estimating the break-even point of an 
agroforestry option. 

Other economic methods: modelling 

Once the above processes are quantified, there are a few additional 
methods that can be used for simulating the profitability, adoptability and 
biophysical consequences of agroforestry practices. These methods all 
entail some form of modelling of the economic processes just discussed, 
in relation either to biophysical processes or to various social and cultural 
variables. Most of them have been developed very recently and have been 
validated in only one set of circumstances. They are mentioned here only 
for reference purposes as they lie outside the scope of this book. 

An example of a bioeconomic model, that is, a model that integrates 
economic and biophysical variables and processes, is provided by Grist et 
al. (1999), who used SCUAF, an agroforestry biophysical model, in 
combination with economic spreadsheets. Another example can be found 
in Shepherd and Soule (1998), who have built a farm simulation model 
which simulates the effects of different land-use management strategies, 
including agroforestry options, on soil processes and nutrient-limited plant 
production and on profitability. 

Adesina et al. (2000) provide an example of a model (Logit) that 
explains actual farmers' decisions to adopt alley farming on the basis of a 
number of explanatory variables. These variables reflect the land tenure 
system, the socioeconomic characteristics of farmers (e.g. level of 
education, membership in farmers' association) and village-specific 

32 A.-M.N.Izac 

characteristics (e.g. land pressure, erosion index and importance of 
livestock in the village). 

2.5 Landscape and Global Scales: Soil Fertility and 
Agroforestry Trees as Part of Natural Capital 

It was shown in Section 2.3 that farmers' decisions at the farm scale not 
only are shaped by variables determined at higher levels than the farm in 
the hierarchy of systems but also have consequences at these higher levels. 
Two concepts from ecological economics are particularly useful in 
understanding these interactions between scales. These are the concepts 
of natural capital and of ecosystem services. 

It has been demonstrated that soil nutrients and trees are part of 
natural capital (Izac, 1997, for soil nutrients; Izac and Sanchez, 2001, for 
agroforestry trees). Natural capital comprises all the natural resources that 
provide useful goods and services for mankind. More specifically, natural 
capital is defined as stocks of resources generated by natural 
biogeochemical processes and solar energy that yield useful flows of 
services and amenities for the present and into the future (Izac, 1997). 
Natural capital generates ecosystem services which are the processes 
ensuring the productivity, integrity, maintenance and resilience of 

The ecosystem services generated by soil nutrients include nutrient 
cycles, soil fertility, plant nutrition and carbon sequestration. The 
ecosystem services generated by agroforestry trees include, for instance, 
erosion control, water cycling, pest and disease control, and biodiversity. 

An important characteristic of ecosystem services is that they occur 
over different spatial and temporal scales (Izac, 1997). Carbon cycling, for 
instance, is a global cycle related to climate change, taking place over the 
atmosphere, soils and ocean -bottom sediments, and doing so over 
relatively long periods of time. At the plot scale, carbon is linked to soil 
fertility. Table 2.3 illustrates the principal ecosystem services of trees at 
different spatial scales. 

Since soil nutrients and trees (natural capital) generate different 
ecosystem services at different spatial scales, different members of society 
will be affected by these ecosystem services. Likewise, the management of 
these ecosystem services and changes in this management affect these 
different groups in society. When the non-monetary benefits of soil fertility 
management through agroforestry were discussed, it was pointed out that 
farmers may be quite indifferent to some of these benefits or ecosystem 
services (e.g. soil biodiversity and carbon sequestration). These benefits 
are, however, valued by other members of society. National societies in 
tropical countries do value sustainability of food production and 

Economic Aspects of Soil Fertility Management 

3 3 

Table 2.3. Principal ecosystem services of agroforestry trees at different scales. 


Ecosystem functions 





Food production 
Nutrient cycling 
Erosion control 
Water cycling 
Genetic diversity 
Microclimate regulation 

Decreased poverty 

Erosion and sedimentation control 

Water cycling 

Refugia, pollination, biological control 

(landscape patches) 

Decreased poverty 

Decreased deforestation and desertification 


Water cycle 

Greenhouse gas regulation 

Climate regulation 


Rural poverty alleviation 

biodiversity, for instance, since most countries have food security and 
biodiversity strategies and the global society does value carbon 
sequestration, as demonstrated by the fact that there is an international 
market for sequestered carbon. 

From an economic perspective, such benefits of the ecosystem services 
of agroforestry systems for soil fertility management are positive 
environmental externalities. These are defined as follows. Investments by 
farmers in agroforestry systems trigger flows of non-monetized benefits 
accruing to different groups in society, called environmental externalities. 
Farmers accrue some of these benefits, largely those related to increased 
risk buffering capacity and sustainable food production (in addition to 
accruing the monetary benefits of agroforestry). National society will 
receive the benefits of decreased rural poverty, watershed protection, 
increased biodiversity, more sustainable agricultural production and 
increased food security. Global society will enjoy increased carbon 
sequestration and biodiversity benefits. 

These various environmental externalities are likely to be highly 
valued by national and global societies because these have a longer 
planning horizon than individual farmers. The relative significance of 

34 A.-M.N.Izac 

ecosystem functions vis-a-vis food and income production is difficult to 
appraise. An attempt at evaluating these functions has, however, been 
made. Costanza et al. (1997) estimated that the global value of 17 ecosystem 
services for 16 key biomes is about US$33 trillion per year. This is 
excluding the worth of the stocks of natural capital in these biomes, as 
stocks of natural capital are too difficult to evaluate from an economic 
perspective. By comparison, the world gross national product (GNP) is 
only worth some US$18 trillion per year and the food produced by all 
croplands amounts to about US$0.13 trillion per year (Costanza et al., 
1997). These estimates give an indication of the order of magnitude of the 
value to society of the ecosystem services generated by natural capital 
compared to the value of agricultural production from croplands and the 
value of total economic production, or GNP. 

The economic method most appropriate for assessing the various 
environmental externalities associated with agroforestry practices is social 
and environmental cost-benefit analysis. This consists of a complex set of 
economic methods (sometimes called shadow pricing) for assessing all non- 
monetary benefits and costs for different groups in society, in addition to 
the assessment of the private costs and benefits described in the previous 
section (see Mishan, 1976, for details). Furthermore, it requires that all 
price distortions due to government or policy failures be redressed as the 
analysis is conducted from the perspective of society as a whole (Mishan, 
1976, provides the best explanation of these various methods, as well as 
of their shortcomings). This is expressed as: 

» U+I t )-lD t +S t ) 

N?V = 2^~ — (2.4) 

1=1 (l + rj 

where: r = social rate of discount; A t = vector of all monetary benefits of 
an agroforestry option during year t, after adjusting prices for relevant 
market failures; I t = vector of all non-monetized benefits received by 
different groups in society during year t, after adjusting prices for relevant 
market failures; D t = vector of all monetary costs of an option, after price 
adjustments for relevant market failures, in year t; S t = vector of all non- 
monetary costs of an option, after price adjustments for relevant market 
failures, in year t. 

The social rate of discount (r s ) is the rate at which society is willing to 
trade present benefits for future benefits. Since society has a longer time 
horizon than individuals, this rate is generally significantly lower (e.g. 5%) 
than the individual rate of discount of relevance to farmers in tropical 

To date, social cost-benefit analysis has not been applied to the 
assessment of agroforestry options, probably because there are some 
currently unresolved methodological challenges. This is, therefore, an area 

Economic Aspects of Soil Fertility Management 

3 5 

for future research. The International Centre for Research in 
Agroforestry, for instance, initiated in 2001 a global research project the 
objective of which is to assess, from an economic perspective, the various 
environmental externalities of key agroforestry practices. 

The ecosystem services and environmental externalities associated 
with agroforestry systems and soil nutrients indicate that what is an optimal 
level of adoption of agroforestry practices from the viewpoint of farmers 
is a suboptimal level of adoption from the perspective of national and 
global society. This situation is represented in Fig. 2.4. The marginal 
(monetary and opportunity) costs of adoption of agroforestry practices are 
compared with the marginal (monetary and non-monetary) benefits of 
adoption received by individual farmers, those received by the national 
society and, finally, those received by the global society. Marginal costs of 
adoption are incurred uniquely by farmers, in the absence of any policy 
geared at implementing a cost-sharing scheme. Marginal benefits of 
agroforestry are higher for the national society than for individual farmers, 
and yet higher for the global society, for the reasons just mentioned. 

Figure 2.4 shows that the level of adoption of agroforestry practices 
that is optimal at the farm scale (from the viewpoint of farmers) is Q p , 
whereas the level of adoption national society regards as necessary is Q N , 
and Q G is the globally optimal adoption level. In the absence of any policy, 
farmers will adopt level Q p , which is suboptimal from society's viewpoint 

Marginal benefits 
(global society) 

Marginal benefits 
(national society) 

Marginal benefits 

Qp Q N Q G 

Level of agroforestry adoption 

Fig. 2.4. Regional and global level trade-offs between the costs and benefits of 
agroforestry systems at a given point in time. 

36 A.-M.N.Izac 

because it does not take into account most of the ecosystem services of 
agroforestry systems. These ecosystem services are therefore unlikely to 
be fully realized. These trade-offs between the individual and societal costs 
and benefits of agroforestry adoption (represented in Fig. 2.4) are 
fundamental economic processes at the regional and global scales of 

The economic processes discussed in this chapter indicate that private 
and social interests concerning soil fertility management and agroforestry 
in tropical countries do not coincide. The extent of agroforestry practices 
voluntarily adopted by farmers (that which is in their own best interest) 
will almost certainly be inferior to that which is socially optimal. It will not 
be optimal or effective or equitable to expect resource-poor small-scale 
farmers in tropical countries to bear the full costs of adoption while 
national and global societies receive significant benefits from this adoption. 

In such circumstances, providing farmers with information and advice 
on soil fertility management and agroforestry practices will definitely be 
insufficient for bringing about socially optimal adoption levels. Some policy 
measures will be needed for social optimality to occur. Discussions of 
possible policy instruments, from cost-sharing schemes, to community- 
based management and carbon offset mechanisms can be found in the 
literature and lie outside the scope of this chapter (see, for instance, Babu 
et al., 1995; Dewees, 1995; Adesina and Coulibaly, 1998; Donovan and 
Casey, 1998). 

The point we wish to emphasize here is that scientists working on the 
design of soil fertility management practices have a moral responsibility 
to ensure that their results are communicated to policy makers in the 
clearest and most effective manner possible. Sufficient information is now 
available about relevant economic processes at play when decisions to 
adopt are made by farmers, that the scientific community can no longer 
continue to argue that all that is needed for optimal levels of adoption to 
take place is good extension work and rational farmers. We now know that 
farmers are indeed rational and that the best extension system in the world 
is not set up to develop the policy instruments needed to bridge the gap 
between individual and societal benefits and between individual costs and 
societal benefits. 

In summary, the key economic processes at the landscape and global 
scales are as follows. 

• Agroforestry trees and soil nutrients are part of natural capital. As 
such, they generate ecosystem services at different spatial scales. 

• These ecosystem services are positive environmental externalities. As 
such, they are non-monetized benefits of adoption that occur off farm 
and accrue to different groups in society (national and global society). 

Economic Aspects of Soil Fertility Management 37 

• These environmental externalities have a high value to national and 
global societies. 

• Farmers do not take many of these environmental externalities into 
consideration in their decisions to adopt because these are off-farm 

• Consequently, optimal levels of adoption from the farmer's perspective 
are lower than the socially optimal levels. 

• Social and environmental cost-benefit analysis is an appropriate 
method for assessing these different processes. This is a new area of 
research in agroforestry. 

• It is essential for scientists to communicate the results of their work to 
policy makers to increase levels of adoption towards socially optimal 

Chapter 3 

Designing Experiments and Analysing 



' World Agroforestry Centre, International Centre for Research in 
Agroforestry (ICRAF), PO Box 30677, Nairobi, Kenya; 2 Institute of 
Soil Science and Soil Geography, University of Bayreuth, 95440 
Bayreuth, Germany; 3 Biological Dynamics of Forestry Fragments 
Project, National Institute for Research in the Amazon (INPA), CP 
478, 69011-970 Manaus, AM, Brazil 

3.1 Synopsis 

Appropriate experimental design and analysis of data are fundamental to 
furthering our understanding of the impact of trees on soil fertility. Although 
there are many standard statistical texts that describe procedures for 
agronomic or forestry trials, the inclusion of one or more tree components 
in an agricultural context tends to increase both the spatial variability and 
the time horizon that have to be considered. This chapter makes reference 
to places where standard designs and analyses are described, but focuses on 
aspects of particular importance in an agroforestry context. 

Given the long-term nature of many agroforestry practices, setting 
appropriate objectives for individual experiments is of vital importance 
but is seldom discussed in standard texts and is often the root cause of 
failure to achieve useful outcomes from agroforestry trials. Here, advice 
on setting objectives is given before consideration of appropriate treatment 
structures, layout and replication. Experiments with tree fallows are given 
special attention. They have become particularly important in soil fertility 
research over the last decade and require careful consideration because 
they generally involve a large number of factors, are subject to interference 
among plots and include treatments with different fallow lengths and 
hence timing and duration. The spatial and temporal heterogeneity in 
soils induced by the presence of a relatively few, large plants that grow 
over several years complicate sampling strategies, so advice is given on the 
most efficient sampling schemes to use to meet different types of exper- 
imental objectives. 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 39 

F.L. Sinclair) 

40 R. Coeefa/. 

Collection of data is of little value unless they are appropriately 
analysed, but objectives for analysis are often modified as research 
proceeds, so a process of exploratory analysis followed by application of 
formal statistical tests is recommended. Particular advice is given on simple 
methods of dealing with repeated measurements in time and space and 
the appropriate statistical software to use in data analysis. 

The final part of the chapter deals in some detail with methods for 
analysis of spatial structure. This is likely to be of increasing importance 
in agroforestry research, where spatial heterogeneity at various scales is 
increasingly embraced as an important component of agroforestry 
practices and landscapes, rather than being seen as something that has to 
be minimized. The techniques for spatial analyses are relatively new and 
have not yet been widely applied in agroforestry contexts. They are 
discussed here in sufficient detail for readers to decide how to approach 
analysis of spatial structure and make reference to the most appropriate 
literature where further details and examples are available. 

3.2 Experimental Objectives, Treatments and Layout 

(R. Coe) 

The design and analysis of experiments for investigating fertility effects of 
agroforestry follow the principles and practices that have been developed 
for field experimentation over the last century or more. These principles 
are summarized here, with references given to more complete descrip- 
tions. Checklists of points to remember when designing an experiment 
maybe useful and are given elsewhere by Coe (1997a, 1999). This chapter 
focuses on some aspects of experiments for investigating impacts of trees 
on soil fertility in which it can be difficult to apply these well-established 

Identifying the objectives of an experiment 

The objectives of an experiment determine every aspect of its design, so 
are a key to a sound design. The objectives must be: 

• Clear: objectives stated vaguely or ambiguously lead to confusion later; 

• Complete: incompletely determined objectives mean that some design 
decisions cannot be made; 

• Relevant: applied research has aims of solving specific problems and 
each experiment must have a clear place in the problem-solving 

• Capable of being met by an experiment: not every research objective can 
be met by an experiment. 

Designing Experiments and Analysing Data 41 

Much of the rest of this book is aimed at helping researchers determine 
what areas of research might be fruitful and how to go about making 
appropriate measurements of key variables. However, this is not the same 
as identifying objectives for individual trials that meet the criteria above 
and which will allow all decisions about the design to be made. Few of the 
texts on experimental design have much to offer on the problem of 
statement of objectives, even though the problems scientists encounter in 
many field trials originate from objectives that are confused or 
contradictory. The classic book by Cochran and Cox (1957) devotes half a 
page to the topic and more recent books do no better, maybe because there 
is little structured advice that can be given. A recent exception is Robinson 
(2000). There are, however, some common problems which can be avoided. 
First, objectives as vague as 'to see what happens' cannot be used to 
design a trial. If the research is still at the stage of having no clear 
hypotheses or quantities to measure then an experiment is not (yet) 
needed. Secondly, careful reading of objectives often reveals them to say 
nothing more than 'the objective of the experiment is to compare 
treatments'. Objectives should be expressed as knowledge gaps to fill, that 
is, hypotheses to test or quantities to estimate, which lead to a definition 
of treatments, rather than starting with the treatments themselves. Most 
experiments have multiple objectives but these have to be mutually 
consistent. A common conflict is between objectives with a technology 
perspective, aimed at evaluating options for farmers, and those aimed at 
understanding processes. These two perspectives can often lead to 
different requirements of treatments and management. For example, Rao 
et al. (1998) describe a trial in which Senna spectabilis was deliberately used, 
rather than a species of greater importance to farmers in the region, 
because the roots were easily distinguished from those of the 
accompanying crops. In technology development trials, a further common 
difficulty is simply that of choosing where to start among the many aspects 
that could be studied. Cooper (1997) describes an approach to thinking 
through this, based on identification of critical constraints, and Franzel et 
al. (2001) discuss the appropriateness of different types of on-farm trials, 
with varying degrees of farmer and researcher control, for meeting 
different objectives. 

Defining experimental treatments 

The key ideas needed to define treatments for an experiment are as 


• Contrasts: most objectives can be reduced to the need to compare two 
or more treatments, either to detect whether a difference exists or to 
measure the size of the difference. Clearly the treatments to be 

42 R. Coeefa/. 

compared or contrasted determine the treatments required in the 

• Controls: control treatments are nothing special, simply the standards 
against which other treatments are compared. They do not have to be 
defined as 'zero input', 'farmer's practice' or similar, but will be chosen 
depending on the objectives. 

• Factorial treatment structure: this can arise naturally from the hypotheses 
or objectives, but can also be used to test several unrelated hypotheses 
within one trial more efficiently than with separate trials. 

• Quantitative levels: when treatments involve different quantities of 
something, then the actual levels required may not be defined by the 
objectives, but guidelines on choosing them are available. 

Mead (1988) gives a thorough practical discussion of statistical issues in 
choice of treatments. 

Defining suitable control treatments should not be difficult if the 
objectives are clear, but there can be conflicts if the trial has multiple 
objectives. Commonly, farming system and biophysical objectives conflict 
and in some cases (at least) two controls are needed. Examples are a 
biophysical control for process-oriented research and a farming system 
control for applied studies, or a forestry and an agricultural control 
because both of these represent well established alternative land uses 
(Dupraz, 1999). The concept of controls is ill-defined and it is clearer to 
simply consider the contrasts needed to meet the objectives and then the 
treatments needed for the defined contrasts. 

Quantitative treatments are common in experiments concerning 
agroforestry and soil fertility - for example, amounts of organic or 
inorganic input and density of trees. For these types of treatments the 
objective usually reduces to estimation of a response curve. Choosing 
treatments involves choosing the range of levels (minimum and 
maximum), the number of different levels, the spacing between the levels 
and the degree of replication of each level. There is a mathematical theory 
of design applicable where a smooth response curve is expected, from 
which the following general rules emerge. 

• Make the range as wide as feasible. Make sure the highest levels are 
well past any optimum. 

• Limit the number of different levels; estimating a straight line needs 
just two. A third will allow its straightness to be checked. More complex 
curves may need four or five levels but it is very rare to find an 
example that needs more different levels than this. 

• Put levels closest together where you think the response will change 
fastest. In the absence of any other information use equally spaced 

Designing Experiments and Analysing Data 43 

When there are two or more quantitative treatments there are important 
alternatives to complete factorial sets of treatments. 

Many agroforestry experiments involve treatments of distinct 
agricultural systems, not just relatively minor variations of the same system. 
This means it can be difficult to define the management of the differing 
plots in a way which does not lead to confounding of treatment effects with 
management variables. A simple example from Kenya involved screening 
seven tree species for their effect on intercropped maize. An eighth 
treatment of monocrop maize was included. As this was a phosphorus- 
deficient site, trees were given a starter dose of phosphorus fertilizer but 
the monocrop was not, resulting in confounding of effects of trees and 
phosphorus fertilization. The design is acceptable for comparison of 
systems, to be assessed perhaps on the basis of profitability, but biophysical 
data on, for example, phosphorus cycling or even crop yield are difficult 
to interpret. Dupraz (1999) gives other examples of this problem, 
including confounding effects of pest and weed control. 

Experimental layout 

Layout of a trial involves choice of locations, definition of the plots, 
characterization of the site, decisions on replication, choice of blocks and 
allocation of treatments to plots. Much of the theory and practice of all 
these developed for agricultural trials is relevant to agroforestry trials. 
There are many books on the topic. Dyke (1988) gives practical field 
advice. Mead (1988) covers the more statistical theory. Notes by Coe 
(1997b, c) and Stern and Adam (1997) are also relevant. 

There are two important differences between typical soil fertility 
experiments involving annual crops and those with agroforestry 
treatments which arise because of the differing time and spatial scales 

• Trees may take several years to produce the effects of interest, so trials 
will usually be in the ground much longer than those with annual 
crops. If a trial of one growing season turns out to have been poorly 
designed or laid out then it can be repeated the following season. This 
is not true for an agroforestry trial in which design faults associated 
with the tree component may not be apparent until years after it was 
planted. Hence, it is worth investing resources in ensuring the layout 
is as sound as possible. In particular it is worth characterizing the site 
to identify heterogeneity, which can then be allowed for by a 
combination of blocking, avoidance and increased replication (see also 
Section 3.6). Another implication of the long time horizon of 
agroforestry experiments is that they will inevitably be used for 


R. Coe ef al. 

purposes other than those for which they were originally planned. It 
is therefore worth building some flexibility into the design. This means 
keeping treatments simple, using large plots that can be split later and 
including extra replicates that may be modified as needed (for 
example, by destructive harvest). 
• Trees are typically larger than annual crops, which also has 
implications for the layout of experiments. Most important is the 
avoidance of interference between neighbouring plots. Above-ground 
interference occurs through such effects as shading, windbreak effects 
and lateral transport of tree litter by wind. These may not be easy to 
control but at least can be seen and perhaps avoided. Below-ground 
interference occurs as tree roots extend beyond their nominal plot and 
capture resources from surrounding areas, either from other plots or 
from outside the experiment. In many trials it is clear that this has 
resulted in biased treatment comparisons. If a plot with trees is 
adjacent to a crop-only plot then the tree plot may capture extra 
resources and the crop plot may be deprived of resources relative to 
a situation in which both are grown in very large plots, resulting in a 
biased treatment comparison. The options for avoiding the bias are 
either to leave very large border areas or to physically prevent 

It is difficult to judge the size of borders required in any situation; this 
clearly depends on the characteristics of the species involved, the site and 
the length of time the trees will grow. Van Noordwijk et al. (1996) discuss 

Fig. 3.1. Lateral root from a Sesbania sesban fallow plot (on the right) grown into a 
crop plot (Cajanus cajan) in about 4 months. 

Designing Experiments and Analysing Data 45 

factors affecting tree root growth in agroforestry systems and make the 
point that it is very difficult to give general guidelines for the lateral extent 
of roots. Certainly, in dry areas, extensions of over 40 m have been 
recorded. Figure 3.1 shows a lateral root of Sesbania sesban known to have 
extended over 6 m in 4 months. 

Physical prevention means installation of below-ground barriers or 
regular root pruning between plots. Barriers are of limited value as roots 
can quickly grow under them and up into adjacent plots (Rao and 
Govidarajan, 1996). Pruning below ground is expensive as trenches have 
to be re-opened each time it is done, possibly four times a year. Trenches 
cannot usually be left unfilled between primings because of their impact 
on water movement and the likelihood that they will behave similarly to 
solid barriers, with roots growing under them. A third alternative that has 
been suggested is to lay out trials so that plots are not adjacent. This will 
probably decrease the precision of the experiment but, more importantly, 
will not remove the small plot bias as trees will still be capturing resources 
from outside their nominal area. In fallow experiments it is common to 
use large plots with large borders for the fallow phase of an experiment, 
then to split the plots for further treatment factors during the cropping 
phase. Remember that the original borders may still have to be left as they 
may be untypical even after the trees have been removed. 

An alternative approach recognizes that, within many farming systems, 
agroforestry will be used in small patches in a landscape. This landscape 
will be full of edges between agroforestry and other land uses, and the 
lateral interactions between them may be an important component of the 
way the system functions. Indeed, this agricultural landscape mosaic with 
trees has been defined as agroforestry (Leakey, 1996). The implication for 
research is that interactions should not be removed from trials but 
incorporated into the objectives (van Noordwijk, 1999). Designing trials 
with a system focus using this notion is feasible, though challenging. The 
problems are likely to be both in selecting useful objectives from the myriad 
of possible topics, and installing, managing and monitoring trials of the 
size required. The alternative is to measure lateral flows in simpler 
experiments and then attempt to model the landscape level effects of 
agroforestry systems. 

The longer time scale and larger plots needed in typical agroforestry 
trials, compared with annual crops, can make trials with farmers difficult. 
Farmers have to commit relatively large areas of land for long periods. The 
areas required mean that often each farmer can only test a few treatments 
and the ideas of incomplete blocks are needed to find effective designs. The 
long time periods also mean that both drop-outs and modifications by 
farmers are common, resulting in more replicates being needed. 

46 R. Coeefa/. 

Number of replicates 

The number of replicates required in an experiment is straightforward to 
estimate in theory, but can be difficult in practice as information required 
is unknown before the trial starts. Replication serves three purposes in an 

• to allow estimation of the random variation, and hence of the precision 
of contrasts of interest; 

• to increase the precision of estimating important quantities, the 
average of a larger sample being more precise than that of a smaller 
one; and 

• as insurance against disasters that lead to loss of parts of the trial. 

The number of replicates needed to satisfy each of these purposes has to 
be looked at. 

The quality of the estimate of a variance depends on the degrees of 
freedom associated with it. For a simple randomized block design the 
variance relevant to estimation of treatment differences is the residual or 
error variance. The degrees of freedom in the residual line of the analysis 
of variance table indicates how well that variance will be estimated from the 
design. Experimenters should aim for residual degrees of freedom to be at 
least ten, though one or two less than this may be acceptable. Residual 
degrees of freedom less than six are hopeless. For more complex designs 
which have several different layers, such as split-plot or multisite designs, 
there is more than one variance relevant to estimation of treatment effects, 
and each has to have sufficient degrees of freedom. The researcher in these 
circumstances has to be able to produce an outline analysis of variance for 
the design and check that degrees of freedom for each relevant error line 
are adequate. Some software can produce these outline analyses, but any 
software can be forced to do so by putting in dummy data, as the degrees 
of freedom depend only on the design, not on the data values. In 
agroforestry trials it is not unusual for the degrees of freedom to be different 
for different variables. For example, a trial that compares monocrop maize 
with maize intercropped with trees has more plots with crop data than with 
tree data. The analyses of variance for tree and crop variables will, 
therefore, have different degrees of freedom, so both need checking. 

Estimation of replication required for the second purpose, controlling 
precision, requires experimenters to have an idea of: (i) the size of relevant 
error variances likely to occur in their trial, and (ii) the precision they 
actually require. The first of these has to be estimated from previous 
experience with other trials of a similar type in a similar environment. 
Although every trial should be different from all predecessors, there are 
few experiments that are so novel that it is not possible to get some idea 
of the expected levels of variation. The variation may be expressed as a 

Designing Experiments and Analysing Data 47 

variance (o 2 ) or independently of the units of measurement as a coefficient 
of variation (cv), which is simply the standard deviation divided by the 
mean. The precision required is most usefully expressed as the standard 
error or width (w) of confidence interval needed for critical quantities. 
Alternatively, it may be expressed as the size of effect that the trial needs 
to detect as significant. Formulae for the number of replicates may then 
be deduced. For example, for an approximate 95% confidence interval of 
width w for the difference between two means in a randomized block 
design, the required number of replicates (r) is 

r = 32— (3-1) 


Software is available for producing formulae for other designs and other 
specifications of the problem (Thomas and Krebs, 1997). 

For the third reason for replicating, insurance, it is harder to judge 
the number required. If an experiment has a short lifespan (say one 
season) then there is little justification for insurance, unless perhaps under 
farmer management where there is some uncertainty about the likelihood 
of individual farmers adhering to important aspects of a protocol. 
However, if a long-term trial is in a situation where it might be at risk (say 
from fire) then inclusion of a few extra replicates may be judged 
worthwhile, particularly given the other reasons for building some 
redundancy into the initial design, mentioned above. 

The situation for the simplest on-farm trials, designed to estimate the 
mean difference between two treatments, is no different from that for on- 
station trials, and the same information is required to estimate sample sizes. 
Variation at all levels tends to be higher, so numbers of replicates needed 
are often large. The large plots needed in many agroforestry trials mean 
that it is often only feasible to have one replicate per farm, and farms 
should be treated as blocks. If it is only possible to have less than one 
replicate per farm then incomplete block designs are useful, with an 
implication for the number of replicates needed. The statistical principles 
are described well by Mead (1988), and software is available to help find 
suitable designs. 

Many on-farm trials have more complex objectives than simply 
comparing two treatment means, but are concerned with performance of 
agroforestry in different farm and landscape situations. Very often there 
is a hierarchy of levels in the design from individual plants at the lowest 
level to communities or watersheds at a higher level. At each of these levels 
there are design questions to resolve concerning the number of units, the 
selection of those units and the treatments or interventions to be compared 
on those units (Table 3.1). The same ideas are needed for determining the 
number of replicates in these more complex designs as in simpler ones. 


R. Coe ef al. 

Table 3.1. Hierarchy of levels and questions to be answered in design of 
on-farm experiments. 


How Which Who What What is 

many? ones? decides? intervention? measured? 


Landscape position 





3.3 Fallow Experiments (R. Coe) 

The principles underlying the design and layout of fallow experiments are 
no different from those for other experiments, the key to a sound design 
being very clear and exact objectives for the trial. However, fallow 
experiments can give rise to particular difficulties because of the potentially 
large number of factors to investigate, the potential for plot to plot 
interference and the fact that time may be involved as a treatment. 

If the experiment is system focused, aimed at evaluating the effect of 
management factors, there are questions concerning fallow establishment, 
management, clearing, post-fallow cropping and starting a second fallow 
cycle. Techniques similar to those described in Tripp and Wooley (1989) 
can be used to prioritize factors for investigation. The idea is to choose 
criteria such as ease and cost of doing the research, the extent to which 
the factor is critical to adoption of the technology, the length of time the 
research will take and the chance that it is successful. The various factors 
that could be investigated are then scored on each of the criteria to 
produce a ranking. The number of treatment combinations can sometimes 
be reduced by assumptions such as, for a given location and species, that 



Fig. 3.2. Schema of soil fertility development during cropping and fallow cycles. 

Designing Experiments and Analysing Data 49 

Table 3.2. Design of a fallow experiment comparing fallow lengths of one, two 
and three seasons. 

Treatment 1 2 3 4 

1 F C 

2 F F C 

3 F F F C 

F, fallow; C, crop. 

there is a direct relationship between the fertility effect of the fallow and 
the fallow biomass (Mafongoya and Dzowela, 1999). When the trial has a 
process focus, the process objectives will determine treatments. Note in 
particular that species for fallows can often be chosen in a way that gives 
the trial much more value than a simple comparison of a range of likely 
species. An example is the comparison of four species deliberately chosen 
to evaluate the two characteristics of deep-rooted vs. shallow-rooted and 
nitrogen fixing vs. non-fixing. The control treatments might be a natural 
fallow, continuous cropping, a herbaceous fallow or even bare ground 
(Hartemink et ai, 1996), the choice, of course, depending on the exact 
objectives (see Section 3.2). 

Much applied research on fallows is concerned with factors associated 
with the rate at which they restore fertility and the rate at which crops 
reduce it - the shape and slopes of the curves in Fig. 3.2. 

Investigation of fertility recovery during the fallow phase requires 
comparison of fallows of different lengths. A design to look at fallows of 
one, two and three seasons may involve treatments as shown in Table 3.2. 

The crops in seasons 2, 3 and 4 will reflect the changes in soil fertility 
following one, two and three seasons of fallow. However, the expected 
pattern (Fig. 3.2) will be confounded with the season-to-season fluctuations 
in weather and disease patterns. For this reason the phased entry design 
of Table 3.3 is preferable. 

Table 3.3. Design of a fallow experiment comparing fallow lengths of one, two 
and three seasons (version 2). 

Treatment 1 2 3 4 

1 F C 

2 F F C 

3 F F F C 

F, fallow; C, crop. 

50 R. Coeefa/. 

Now the effects of differing fallow lengths are assessed by comparison 
within a single season. However, the design raises the two following 

• What should be put in the cells marked * ? At the start of the first 
season the whole site is presumably in the state required for the 
experiment. To avoid confounding fallow length with other factors we 
need the soil in treatment 2 plots to be the same as this at the start of 
the second season, and those of treatment 1 plots at the start of the 
third season. There is no practice that can guarantee this. If the trial 
is concerned with farming systems then it is common for the initial 
condition to be that state of degradation in which a farmer would 
abandon cropping and start a fallow phase. It is often assumed that 
one or two further seasons of cropping on these degraded plots will 
not substantially change the soil further, so the design used is as in 
Table 3.4. 

• The designs in Tables 3.3 and 3.4 are appropriate if the seasonal 
effects (such as weather) apply to the crop but not the fallow. Although 
it is likely that in general a crop will be more sensitive to seasonal 
differences than a fallow, this need not be the case. Suppose the third 
season was a very good one for fallow establishment, but that the 
second season was not. This difference may be reflected in the 
following crops and be confounded with the comparison between the 
two seasons. A way around this is to repeat the sequences in more than 
one season (Table 3.5). It is clear that this design can quickly grow to 
something unmanageable, both because of the number of plots and 
the time involved. Of course there is no guarantee that the seasons 
sampled with just one repeat of the sequences will be sufficient. Sites 
are sometimes substituted for seasons, putting treatments 1, 2 and 3 
down simultaneously in a number of locations. 

When looking at the cropping phase of the sequence there are similar 
considerations. If we want to measure fertility decline over, say, three 

Table 3.4. Design of a fallow experiment comparing fallow lengths of one, two 
and three seasons (version 3). 

























F, fallow; C, crop; D, crop on degraded soil. 

Designing Experiments and Analysing Data 


Table 3.5. Design of a fallow experiment comparing fallow lengths of one, two 
and three seasons, repeated through two different seasons. 















































F, fallow; C, crop; D, crop on degraded soil. 

seasons of cropping following a fallow, the comparisons require a design 
as in Table 3.6. In the fourth season the first, second and third crops 
following a 1 year fallow are compared. 

It is clear that the requirements for comparing differing cropping and 
fallow periods are mutually incompatible. It is not possible to both start 
and end different treatment sequences at the same time if their lengths 
differ. For this reason compromise is necessary, the usual solution being 
to hope that fallows are less sensitive to seasonal variation than crops. The 
fallow plots can be measured to find out whether growth is in fact similar 
in different seasons (survival and biomass at the end of the first season of 
growth would be appropriate for this purpose), leaving open the question 
of what to do if this turns out not to be so. A typical design to study the 
decline in fertility following fallows with durations of one and two seasons 
uses treatments 1-6 in Table 3.7. 

An alternative approach uses the idea that it is possible to define a 
control treatment (perhaps fully fertilized) that responds to weather and 
other season-specific factors but is constant with respect to soil fertility. A 
trial to look at decline in fertility following fallow might then use just 
treatments 3, 6 and 7 from Table 3.7. Crop yield of treatments 3 and 6 

Table 3.6. Design of a fallow experiment comparing cropping cycles of one, two 
and three seasons. 

























F, fallow; C, crop; D, crop on degraded soil. 

52 R. Coe etal. 

Table 3.7. Design of a fallow experiment comparing cropping cycles of one, two 
and three seasons following one or two seasons of fallowing. 

























































F, fallow; C, crop; D, crop on degraded soil; Cont, control. 

during seasons 3, 4 and 5 would be assessed by measuring them relative 
to those in treatment 7. The approach depends on the assumption 
(untestable with the design) that there is a scale on which there is no 
treatment x season interaction, which seems unlikely, particularly when 
considering variables other than crop yield. Note that if treatments 1 to 7 
of Table 3.7 are used then the value of this approach can be assessed by 
comparing the trends derived from the time series with those measured 
in season 5. 

3.4 Measurements and Sampling Designs (C. Schroth, R. Coe) 

Much of this book is concerned with measurement and there are many 
other sources with practical information both on what to measure and how 
to measure it. When planning experiments, it is worth thinking of the 
following three categories of measurements. 

• Measurements that are needed to characterize the site in which the 
trial takes place, but do not vary between treatments. These include, 
for example, soil type and rainfall. 

• Responses specified by the objectives, which might be soil nitrogen 
and plant growth. 

• Measurements needed to understand variability in the key responses, 
which might be soil depth or texture and are often forgotten. They 
can be specified once the nature of the variability becomes apparent 
and field observation suggests some possible causes. 

A special case of the first type of measurements are those that are 
specifically made to show that basic site characteristics do not vary 
significantly between treatments. This is especially important when proper 

Designing Experiments and Analysing Data 53 

replication of treatments is not possible, because an experiment has not 
been established as such, but is rather a comparison of plots that have been 
under different vegetation cover or land use for a certain time (such as 
comparing agroforestry, conventional agriculture and natural forest). Such 
'unplanned' comparisons often provide information that planned 
experiments cannot yield for practical reasons, especially if long time 
periods under a certain land use are needed. However, they always suffer 
from the difficulty of establishing whether the initial soil conditions were 
similar among sites. Sanchez (1987) suggested particle size distribution in 
the profile as the principal criterion for checking the comparability of sites. 
Nearly all measurements in experiments require sampling of some 
sort and strategies for sampling soils are discussed in detail here. There 
are several recent descriptions of soil sampling procedures for agricultural 
(James and Wells, 1990) and environmental purposes (Crepin and 
Johnson, 1993). When adapting such recommendations for agroforestry 
situations, it is essential to be sure about the aim of the study, as this will 
affect decisions about the sampling plan as well as sampling depth and 
timing. There are two key aspects to be clear about. 

• The level at which replication is needed to estimate precision. As an 
example consider plots of an experiment to be sampled to determine 
differences in soil properties under different treatments. Although 
several samples may be taken in each plot, it is not usually necessary 
to estimate the precision of values for each plot. Instead we need values 
from replicate plots to estimate precision of differences between 
treatment means. 

• Whether it is necessary to compare different parts of the plot or field. 
In many agriculture or forestry experiments we do not expect 
different parts of a plot or field to vary systematically - one point is 
equivalent to any other point. This is not the case in typical 
agroforestry plots. Tree effects will not be uniform across a plot; for 
example, the soil properties under a hedgerow might be expected to 
differ from those under the crop between two hedgerows. The 
sampling used will depend on whether assessing this difference is 
required to meet the objectives. 

Composite samples and bulking 

In most cases, the objective of a sampling programme is to obtain average 
values of certain soil characteristics for a given area. In this case, the soil 
from different sampling points is bulked, mixed and a subsample taken 
for analysis. If this is done, then the variation between repeated subsamples 
gives no information about variation in the field. 

54 R. Coeefa/. 

In some cases it is necessary to take measurements on individual 
samples collected in the field. An example is the estimation of the precision 
of a plot average, usually the standard error of a mean of several samples 
in the plot. This information is useful for optimizing sampling schemes 
during preliminary studies but is of no use for the comparison of 
experimental treatments in different plots. Estimating precision is 
straightforward if the samples were collected in a random pattern. 
Systematic sampling does not allow precision to be estimated in a simple 
way, but estimates are available through the use of geostatistical methods 
to analyse the spatial patterns of soil properties in a plot (see Section 3.6). 

Sampling uniform areas 

An area to be sampled (a field or experimental plot) that is uniform, with 
no known or predicted patterns of variation across it, is unusual in 
agroforestry research. Possible examples for uniform areas are the crop- 
only control in an experiment, agricultural plots following a homogeneous 
fallow and perhaps very dense and homogeneous fallows themselves. 

The sampling scheme determines where samples will be located in the 
area. Two schemes for this situation are random and systematic. Random 
sampling ideally means that the coordinates for each sample location are 
selected using random numbers, but in practice the samples are usually 
collected on a zigzag path across the plot (James and Wells, 1990). In 
systematic sampling, the sample sites are selected in a systematic way, 
typically using a grid pattern. For example, a 20 m x 20 m plot could be 
sampled by locating 16 points in a square 5 m x 5 m grid, starting 2.5 m 
from the edge of the plot. 

Random sampling has the advantage that it is, in theory, unbiased, as 
no part of the plot is favoured. However, the practical application of 
random sampling can be biased as data collectors tend to avoid points they 
feel do not look 'typical'. Systematic sampling has the advantage that it is 
unbiased except in the case of the grid points coinciding with some pattern 
in the field. The example above would clearly be inappropriate if trees 
were also planted on a 5 m x 5 m grid. The practical application of 
systematic sampling is simpler than random sampling because the same 
sample positions can be used in every plot and the subjectivity found in 
practical random sampling is removed. Systematic sampling will nearly 
always give estimates of higher precision than random sampling because 
the sampling points are more evenly spread through the area of interest 
(Webster and Oliver, 1990). 

Designing Experiments and Analysing Data 


Sampling non-uniform areas 

Most areas to be sampled vary in a known or predictable way. The patterns 
of variation may be due to factors such as slope, soil depth or weed 
distribution. In agroforestry plots, pronounced variation of soil 
characteristics, litter and root distribution is often caused by the spatial 
arrangement of different trees and crops, and a common problem is how 
to obtain a sample that is representative for the whole plot in this situation. 
A useful approach to this problem is to identify the smallest representative 
unit (SRU), which is the smallest spatial element of which the plot (or 
system) is composed. For example, in a planted fallow with trees at 2 m x 
2 m spacing, the SRU would be a 1 m x 1 m area with a tree in one corner 
(Fig. 3.3). In a system with trees planted in rows between crop fields with 
5 m spacing between trees in the rows and 20 m between rows, the SRU 
would be a 2.5 m x 10 m strip perpendicular to a tree row with a tree in 
one corner if the direction from the tree row is not considered important 
(SRU 1 in Fig. 3.4), or a 2.5 x 20 m strip if there are directional effects of 
slope or prevailing wind that need to be taken into consideration (SRU 2 
in Fig. 3.4). Even in a pure crop field the SRU concept may be useful, for 
example, when a field has been ridged, when a crop is still small and does 
not occupy the soil homogeneously or in most cases when root systems and 
related soil characteristics are studied. 

To be representative for the whole plot or system, a sample would have 
to be representative for one or preferably more SRUs in a plot. To obtain 
a representative sample of litter in an agroforestry plot, an efficient strategy 
can be to collect the whole litter in one or more SRUs per plot, then 
homogenize and subsample. For measuring coarse root biomass, the 
excavation of the soil of one or more SRUs per plot to a certain depth may 




1 m 



Fig. 3.3. Smallest representative unit (SRU) for sampling an area with trees planted 
at a spacing of 2 m x 2 m. 


R. Coe ef al. 


2.5 m 

2.5 m 


Fig. 3.4. Two options for the smallest representative unit (SRU) for sampling a crop 
field between rows of trees that are planted at 5 m x 20 m spacing. The choice 
between the two SRUs depends on whether a significant influence of the direction 
from the tree rows on the variables under study is expected (SRU 2) or not 
expected (SRU 1). 

be necessary. For soil or fine root samples, the individual sampling points 
would have to cover one or more SRUs in a plot in a representative grid 
or other pattern. 

In many cases, the precision of sampling can be increased, and 
additional information on spatial structures obtained, by stratification. This 
simply means dividing the plot into regions thought to be homogeneous, 
and sampling each of these strata. Within each stratum, either a random 
or a systematic sampling plan can be used. For example, in a pasture or 
crop field with scattered trees, the strata to be sampled separately could 
be: (i) the area influenced by the shade and litter of a certain tree species, 
and (ii) the area not directly influenced by the tree. In a hedgerow 
intercropping experiment, the plot could be subdivided into: (i) the area 
under the tree rows, and (ii) the area between the tree rows that is tilled, 
cropped and receives tree mulch. This cropped area could present 
gradients of nutrient availability caused by decreasing nutrient uptake by 
the trees and/or increasing nutrient uptake by the crops with increasing 
tree distance. In this situation, separate sampling of the soil under the trees 
and crops combined with systematic sampling within the cropped area 
(e.g. every 25 cm on a line crossing the area from one hedgerow to the 
other) ensures representative inclusion of soil from all tree distances. 
Stratified systematic sampling plans are usually the method of choice for 
root studies in agroforestry to allow for small-scale patterns in root mass 
and density around individual trees or tree rows. 

Designing Experiments and Analysing Data 


When a stratified plan is used but interest is in the whole plot, care 
must be taken to weight the observations from each stratum appropriately 
(Rao and Coe, 1991). The following sampling design for a 1 m x 1 m tree 
plantation was used by Jama et al. (1998). The plot is divided into three 
strata according to how far each point is from the nearest tree. The strata 
of 0-0.28 m, 0.28-0.48 m and 0.48-0.7 m from a tree represent 25, 50 and 
25%, respectively, of the area (Fig. 3.5). The choices now are: (i) sample 
each stratum, measure the soil property of interest for each and average 
the results using weights of 0.25, 0.5 and 0.25, respectively; or (ii) create 
a representative bulk sample from each stratum in the ratio 1:2:1. A 
sampling scheme with five instead of three tree distances for the same 
situation has been used by Mekonnen et al. (1999). 

Pronounced spatial patterns of soil fertility within a plot, which have 
to be taken into account when defining sampling schemes, result not only 
from the presence of different tree and crop species in a system, but also 
from standard agricultural practices such as fertilizer placement or certain 
tillage methods. Tree crops are commonly fertilized non-uniformly, for 
example, in bands along coffee rows (on sloping land on the upper side 
of the row) or in a circle around the stem of larger, more widely spaced 
trees (as is common for citrus, oil palm and other tree crops). The area 
where the fertilizer is applied is often kept free from ground vegetation 
to reduce nutrient uptake by weeds or cover crops. Over the years, this 

1 m 

1 m 

Fig. 3.5. Schematic diagram illustrating the subdivision of a plot with trees planted 
at 1 m x 1 m spacing into three sampling strata (reproduced with permission from 
Jama ef a/., 1 998). Note that the 1 m 2 square around the tree contains four smallest 
representative units and that for most research objectives it is better to collect one 
or two samples per distance per tree from several trees in a plot than many 
samples from a single tree. 

R. Coe ef al. 

practice can lead to pronounced differences in chemical soil characteristics 
between fertilized (and weeded) and unfertilized (vegetation-covered) 
areas in a plot. For example, available soil phosphorus (Mehlich 3) at 0-10 
cm soil depth in an Amazonian oil palm (Elaeis guineensis) plantation was 
165 mg kg- 1 at 1 m stem distance, but only 4 mg kg- 1 at 2.5 m stem distance 
(Schroth et al., 2000a). In such situations, it is best to sample and analyse 
the fertilized and the unfertilized soil separately and to complement the 
analysis with a root distribution study to obtain an idea of the relative 
importance of these soils of differing fertility for plant nutrition. This 
example also illustrates how sequential samplings of the same plot could 
erroneously detect changes in soil fertility by collecting samples at slightly 
different positions if spatial fertility patterns are not taken into account. 

Tillage practices that lead to pronounced spatial fertility patterns 
within a field include ridging, a common system not only on sloping land, 
but also on level areas, for example in West Africa. With a specially 
designed plough or a hoe, a strip of soil is turned and moved sideways, 
where it covers a soil strip of similar width and forms the ridge. Topsoil 
and weeds are concentrated in the ridges, on which the crop is sown. 
Microerosion progressively takes fine materials from the ridges into the 
furrows during the growing season. If sample collection before the ridging 
is not possible, sampling halfway up the ridges has been recommended 
(James and Wells, 1990), but it is often better to collect and analyse two 
separate samples from the ridges and the furrows, especially if different 
crops are grown in these two positions. 

Comparing parts of a plot 

In agroforestry experiments, plots are not only subdivided to avoid 
problems with sampling heterogeneous areas. In many cases, the analysis 
of the spatial patterns of soil properties that are caused by the different 
plant species is of considerable interest itself, as it may provide information 
on the different effects of trees and crops on the soil and so on tree-crop 
interactions. Examples would be systematic changes in soil nitrogen or soil 
water content with increasing tree distance in a crop field. Figure 3.6 shows 
a plot with a single row of trees down the centre with crops on either side 
(as would be appropriate for a shelterbelt). To measure the spatial 
variation of soil properties caused by the trees, samples have to be collected 
at different distances from the tree line. The principles are the same as for 
choosing quantitative levels of a treatment factor (see Section 3.2). If a 
simple response to distance is to be estimated, then it is better to use many 
samples at a few distances rather than few samples at each of many 
different distances. Typically, measurements will be taken in or close to 
the tree row (maximum tree effect), at a large distance (no tree effect) and 

Designing Experiments and Analysing Data 



Fig. 3.6. Sampling scheme in transects in a plot with a single tree row and crops. 

at one or two intermediate distances. Soil properties usually change most 
rapidly close to the trees, and it is thus advisable to collect samples at 
smaller distance intervals here and increase the intervals with increasing 
tree distance. Precision for comparing distances will be maximized if the 
samples fall in a number of transects, as in Fig. 3.6. 

In certain cases, measuring a representative value for a whole plot or 
system may not be the main objective of a sampling strategy. It may be 
more important, for example, to detect extreme values within a plot. When 
tree species that differ in their effects on the soil are mixed in a complex 
system, such as a forest garden, it woidd be expected that the effect of each 
tree is most pronounced close to the tree and decreases with increasing 
distance from the tree, where it also overlaps increasingly with the effects 
of neighbouring trees. In this situation, the most sensitive sampling 
strategy for the detection of specific tree effects on soil fertility is the 
collection of soil samples taken from very close to individual trees of the 
different tree species for comparison with controls taken at points at a 
maximum distance from all trees, which are assumed to have the smallest 
influence from trees possible given the configuration of the system. The 
resulting single-tree effects on soil fertility cannot be quantitatively scaled 
up to the whole plot unless extensive sampling is undertaken to 
characterize the gradients and geostatistical techniques are employed (see 
Section 3.6), but they do give valuable qualitative information concerning 
the way a certain species and its management might influence soil 
conditions, which is often of more immediate practical use than 
quantitative mean data per hectare (see Schroth et al., 2001b, for a review 
of agroforestry applications). In parklands and similar systems with 
naturally regenerated trees in irregidar arrangements, care must be taken 
in the interpretation of observed fertility differences between tree and no- 
tree positions, as the trees may have established themselves on fertile spots, 
such as termite mounds or depressions receiving run-on water and eroded 
nutrients. The possibility of such initial fertility differences between 
positions may be excluded when the trees have been planted at a regidar 
spacing, as in tree crop plantations. 

60 R. Coeefa/. 

Sampling depth 

Soil sampling for agricultural purposes often includes only the topsoil, that 
is, the plough layer or a shallow surface horizon in no-till fields (James and 
Wells, 1990). In forestry, deeper sampling sometimes gives better 
correlations with plant growth or fertilizer response, for example 60 cm 
for phosphorus and 150 cm for potassium in different pine stands as 
reported by Comerford et al. (1984). For nitrate, sampling to 120 or even 
180 cm depth has been recommended even for annual crops (Comerford 
et al, 1984; James and Wells, 1990). 

In agroforestry, sampling of the top 10 or 20 cm of soil may sometimes 
be adequate when treatments that are unlikely to affect the subsoil are being 
studied, such as the short-term effects of mulching, but it is inadequate for 
studies of competition and complementarity of nutrient use between 
different species. As a rule, the decision about sampling depth should be 
based on an at least qualitative analysis of the vertical root distribution of 
the relevant plant species (see Section 12.2). If this analysis reveals that one 
species (the crop) has relevant amounts of roots to 60 cm soil depth and the 
other species (the tree) to 100 cm depth, it is important to collect a separate 
sample from the 60-100 cm depth as the water and nutrients here may 
contribute to complementarity between the species (see Box 5.2, p. 100). If 
the tree penetrates a hardened subsoil horizon but the crop does not, the 
hardened horizon and underlying horizons should be included in the 
sampling, because increased access to the subsoil for the crop following root 
channels of the tree could be an important benefit of the association. If 
nutrient recycling from the subsoil is to be assessed, a preliminary 
evaluation of the nutrient distribution in the soil profile is useful. 

It is common practice to take fewer samples from depth than from 
surface layers to reduce the sampling effort. However, few samples will 
lead to imprecise estimates, unless the variability of the investigated 
variable also decreases in the subsoil. 

Sampling time and frequency 

Certain soil characteristics change both during the year and between years, 
and sampling time has thus to be selected carefully in relation to the 
objectives of the study. Seasonal fluctuations have been reported for labile 
fractions of carbon, nitrogen and phosphorus as well as microbial soil 
characteristics in response to moisture changes, crop phenology and 
management (Conti et al. , 1992; Mazzarino et al. , 1993; Campo etal., 1998). 
Soil test readings for phosphorus and potassium probably decrease during 
the cropping season, whereas soil acidity increases in acid, but not in 
alkaline soils (James and Wells, 1990). In long-term studies, samples 

Designing Experiments and Analysing Data 61 

repeated in different years should be collected at the same time of the year 
relative to weather and crop development to be fully comparable (e.g. '2 
weeks after crop emergence' or 'as soon as 50 mm cumulative rain has 
fallen after 1 September'). Short-term fluctuations of nutrient availability 
in response to management measures such as fertilization and mulching 
can also be studied by collecting and analysing the soil solution at 
appropriate intervals (see Section 7.2). 

3.5 Analysing the Data (R. Coe) 

There are many resources that describe standard statistical procedures for 
analysing data from experiments. There are, however, few books that 
actually describe the process of analysing data from field trials, so this is 
discussed briefly here. 

The first step is to define objectives for analysis. The difficulty many 
scientists face with statistical analysis originates in part from the analysis 
objectives not being carefully thought through. Defining the objectives of 
analysis will involve: 

• identifying the exact comparisons to be made or relationships to be 

• determining the exact data that are needed to make the comparisons 
(for example, are comparisons of yield needed each season or totalled 
over all seasons?); and 

• designing the tables and graphs that will be used to present the results. 

The objectives of analysis are determined by the objectives of the trial. 
However, the analysis objectives are distinct from trial objectives in that: 

• the objectives of the trial may well have been stated in a rather vague 

• new objectives will have been developed as the trial progresses, 
resulting from observations made; 

• the objectives set out in the original protocol may well have other, 
unstated, objectives added if these can usefully be met with the data 

• it might not be possible to meet all the original objectives of the trial, 
either because the trial design does not allow this, or because 
something unexpected has happened to prevent it; and 

• the objectives of the analysis will evolve as the analysis proceeds. 

The second step requires preparation of the data. Almost always there 
is considerable work to do to turn raw field records into data files suitable 
for statistical analysis. Conversions and calibrations have to be completed. 
New variables have to be constructed. Data have to be summarized to the 

62 R. Coeefa/. 

appropriate space and time scale, and so on. The ease with which this is 
done depends largely on how the original records were organized. 
Statistical Services Centre (2000) is a good guide to optimizing this. 

The third step is exploratory analysis of the data that aims at finding 
summary tables and graphs that meet the objectives. Well-constructed 
tables and graphs will make the patterns and relationships in the data clear 
and should make tentative conclusions obvious. The other aim of 
exploratory analysis is to reveal any unexpected observations or patterns 
in the data that might impact either on the conclusions or the way the 
analysis proceeds. 

Next comes the confirmatory analysis. This is where formal statistical 
techniques are relevant and is the step widely described in books and 
taught in courses. The formal analysis aims to: 

• confirm that the patterns and relationships noted in the exploratory 
analysis are consistent with real or repeatable effects, and are unlikely 
to be due to noise; 

• estimate the precision of important quantities emerging from the 
analysis; and 

• increase the precision of these estimates by finding parsimonious 
statistical models which describe the data, do not contain any 
unimportant terms and manage to explain much of the variation. 

The final step involves interpretation and reporting. Iteration between 
the various steps of analysis is nearly always needed. 

Data from agroforestry-fertility field experiments are not inherently 
different from data from other trials, so these general methods usually 
apply. However, there are some characteristics of data from agroforestry 
trials which can make it confusing to apply general methods. 

Multiple outputs 

In system trials aimed at evaluating the productivity of alternative systems 
there will often be multiple products (tree and crop products, for 
example). The options for analysing these are to look at them separately, 
to investigate the relationship between them and to combine them into a 
single index of production. The range of methods developed for analysis 
of intercropping trials are appropriate (Federer, 1993, 1998). 

Measurements repeated in time 

The time scale of agroforestry experiments means that measurements will 
almost inevitably be taken repeatedly on the same variable at several times. 

Designing Experiments and Analysing Data 63 

These may be within one season, for example to look at nitrogen dynamics 
during a cropping phase, or across seasons to look at long-term 
development of systems or response to different weather. The statistical 
problem associated with repeated measures arises from the fact that 
observations on the same units at different times cannot be expected to be 
independent, with correlations between observations likely to be larger 
the smaller the time interval between them. The most common statistical 
procedures, such as those based on analysis of variance (ANOVA) and 
regression, require independence of observations. A large group of 
techniques under the general heading oirepeated measures analysis has been 
developed to handle this problem. These techniques range from simple 
and approximate adjustments of standard analyses to statistical models 
that describe the correlations in detail. They are described in many 
textbooks and appear in standard software. However they can be difficult 
to apply, particularly if the times of measurement are irregular or different 
for different treatments. 

There is one approach which is simple to apply and produces analyses 
which are well tuned to meeting objectives. The idea is to turn the repeated 
measurements on each experimental unit into a single value for each unit, 
that value being chosen to represent a quantity required by the analysis 
objectives. The steps are as follows. 

1. Explore the data (usually by plotting the results for each treatment 
against time) in order to identify a pattern or effect that is of interest 
and needs formal statistical analysis. 

2. Choose a number that describes the effect of interest. Examples 

• the total biomass produced over 5 years; 

• the difference in biomass produced in a wet season and a dry 

• the change in soil nitrogen between crop planting and leaf stage 6 
minus crop nitrogen uptake, representing nitrogen losses in the 
early part of a season; 

• the time until soil carbon reaches some threshold level, based on 
interpolation of annual measurements; 

• the rate parameter k of a decomposition curve fitted to biomass 
decomposition data. 

These examples show that the summary can be something as simple 
as a total or difference, or as complex as the parameters of a non-linear 
curve fitted to trends. The important point is that the number 
describes the phenomenon of interest. 

3 . Calculate the number for each experimental unit or plot of the original 
design and analyse for treatment differences using the ANOVA 
dictated by the design. 

64 R. Coeefa/. 

4. Repeat the procedure for as many different summary numbers as are 
needed to meet all the objectives. 

Measurements repeated in space 

Repeated measurements also occur in space, as different parts of a plot 
are measured with the purpose of comparing, but they do not represent 
separately randomized treatments. In many soils experiments, 
measurements are taken at different depths. Agroforestry experiments 
often have the added complication that different positions within a plot 
are not equivalent, for example different distances from a line of trees. 
The statistical problem associated with these repeated measures in space 
is identical to that posed by repeated measures in time - the observations 
from different parts of a plot cannot be expected to be independent. The 
approaches for dealing with the problem are also similar, though the 
statistical modelling of correlation structures in space can be rather harder 
than the equivalent in time. The simple approach based on summary 
quantities is also applicable to repeated measurements in space. For 
example, suppose tree and crop root length densities are measured at five 
different distances from a tree line and at eight different depths within 
each plot of an experiment. Depending on the exact objectives, examples 
of suitable summary values to be calculated for each plot and subjected to 
formal analysis might be: 

1. Proportion of tree roots below the crop root layer; 

2. Total tree roots per metre of tree line; 

3. Root length density of crops in the zone 0-50 cm deep and less than 
3 m horizontally from the tree line; 

4. Parameters^, k and c in the models e-ap(-k(h 2 +cv 2 ) 05 ) that describes 
the change in tree root length density with horizontal (h) and vertical 
(v) distance from the base of the trees. 

It is likely that the summary chosen cannot be calculated in the same 
way for all treatments. Suppose the examples above come from a trial that 
compared four different tree species grown in a crop, together with a crop- 
only treatment with no trees, a total of five treatments. Summary (1) is 
simply not defined for the crop-only treatment. When analysing this 
variable there are only four treatments to consider. Summary (2) makes 
sense for the crop-only treatment but has the fixed value of zero. Since 
there is no uncertainty in this treatment mean, it is not correct to include 
these zeros in the analysis along with other values. When comparing a 
mean from a tree plot with that of the control we are comparing an 
uncertain non-zero mean with a known value of zero. For summary (3) we 
may well want to compare the value from tree plots with that from crop- 

Designing Experiments and Analysing Data 65 

only plots. The value for the crop-only plot will be calculated from all 
measurements 0-50 cm deep in the plot, as different locations horizontally 
are all equivalent. The fact that the summary is calculated in different ways 
in plots of different treatments does not affect the formal analysis, apart 
perhaps from inducing some non-constant variance, which can be allowed 
for by suitable weighting. 

Software for experimental design 

No software will design experiments for you. However, there are several 
components of experimental design for which software can help. Software 
to help estimate the number of replicates or sample sizes needed is 
reviewed by Thomas and Krebs (1997). Their online paper also gives links 
to providers of commercial and free software. 

Design of efficient incomplete block designs, neighbour designs and 
other advanced designs that allow for some site heterogeneity is assisted 
by the CycDesign software from the Commonwealth Scientific and 
Industrial Research Organisation (CSIRO), Australia ( 
tigr/software/cycdesign). Similar facilities are available with the Genstat 
programme ( There are many software products available 
aimed at helping to design industrial experiments. These are mainly 
concerned with selection of efficient treatment combinations from very 
large multifactor sets and have little application in agroforestry research. 

Software for statistical analysis 

There are hundreds of products to help in statistical analysis of data from 
experiments. The number around is now so large that there seems to be 
no systematic and reasonably complete list or directory to them. Internet 
sites of many statistical organizations contain links to software sites. One 
extensive listing that may be useful is found at 
uk/cti/links_stats/software.html#packages. Many of the products listed 
here are accompanied by independent reviews. In addition to statistics 
packages there are facilities for certain statistical manipulations in many 
other types of software - spreadsheets, databases, even graphics software. 
With such a large choice it is difficult to make recommendations, though 
some general comments may be useful. 

First, it is sensible to use a dedicated statistics software package for 
statistical calculations. An excellent article explaining the limitations of the 
spreadsheet Excel for statistics, for example, is found at 
uk/ssc/dfid/booklets/topxfs.html. Similar points could be made about trying 
to do anything other than the simplest statistics with, say, database or 
graphics systems. 

66 R. Coe ef al. 

Secondly, there is no statistical software that meets all the requirements 
of any analyst. For example, S-Plus ( has many 
of the more recent developments in statistical methodology incorporated, 
but has a steep learning curve and is not easy for beginners to start using. 
Minitab (, on the other hand, is ideally suited to 
beginners but may well not be able to analyse some of the more complex 
designs found in agroforestry research. 

Generally you get what you pay for. It is unlikely that you will find 
free or shareware products that allow you to complete effective analyses. 
However, cost is not a guarantee of quality or appropriateness for analysis 
of experiments, and software which is sold as a general statistical analysis 
system may not be particularly suited to analysis of experiments as opposed 
to surveys, for example, SPSS ( Possibly the only freely 
available statistical software that does make a suitable introductory package 
for experimenters is instat (, which 
comes with extensive high quality documentation and examples. 

sas ( is often considered as the standard against which 
other statistical software is judged. It certainly has very extensive facilities 
and is widely used. Because of this, analyses using sas are described in 
many books. However, there are some disadvantages. It can be difficult to 
carry out some non-standard analyses or manipulations that are con- 
ceptually straightforward. More importantly, sas has developed a system 
for maintaining, integrating and delivering information from a wide range 
of sources. The applications development and database facilities have 
become very advanced, but will probably not be exploited by the 
agroforestry researcher, who ends up having to pay for software that will 
not be used. The decision on whether or not to invest in sas will generally 
be an organizational rather than an individual one. 

Many less-common analyses will require special software. Software for 
spatial statistics is reviewed in Section 3.6. Other examples of possible 
relevance are mim ( used to unravel complex 
multivariate dependencies, or pc-ord ( for analysis 
of species composition data. 

Given the great choice in software available and the range of tasks the 
analyst of agroforestry trials has to carry out, the only strategy that can be 
recommended is to acquire one of the well known and general statistical 
analysis systems and become proficient at using it, but recognize that 
additional software may be needed for particular jobs. The choice of 
which main package to use will often depend on who else is using it and 
what other tasks, beyond analysis of agroforestry experiments, it will 
be used for. An example of software which allows straightforward analysis 
to be completed easily, leads naturally to more complex analyses and 
overall maximizes the chance of doing a sound analysis is genstat 

Designing Experiments and Analysing Data 67 

3.6 Spatial Structure and Its Analysis (B. Huwe) 

There are several aspects to problems associated with spatial heterogeneity 
or variability in agricultural and agroforestry systems. On the one hand, 
field variability is regarded as a major obstacle in identifying parameter 
impacts or relationships between system components. High variability 
obliges the experimenter to increase the number of replicates in the 
experimental design or to subdivide the study area into several, separately 
sampled strata, which increases the number of samples and the complexity 
of the data analysis (see Section 3.4). On the other hand, the spatial 
variabilities of soil, vegetation and other properties are important system 
characteristics and thus of interest themselves. For example, the detection 
of spatial correlation between the distribution of plant species or their root 
systems and soil properties may be a first step in the analysis of plant-soil 
interactions (Adderley et al., 1997; Mekonnen et al., 1999). Further, in 
many applications extreme values can be more important than mean 
values, such as leaching in macropore systems, influenced by tree roots or 
macrofauna (Beven, 1991), or denitrification in 'hot spots' such as local 
clay enrichments or soil aggregates with reduced aeration (Clemens et al., 
1999). Spatial patterns and probability distributions are essential in studies 
where uncertainty analyses are involved, such as the risk of yield failure 
in drought years as influenced by variable soil texture or soil depth in a 
landscape, groundwater pollution with pesticides or nitrate (Soutter and 
Pannatier, 1996; Wade et al., 1996), or N 2 emissions into the atmosphere 
(Velthofrfo/., 1996; Clemens et al, 1999). 

This section focuses mainly on geostatistical concepts, which means 
spatial analyses and predictions, and their potential applications in 
agroforestry research. Limitations of geostatistical methods are pointed 
out, and some alternatives are mentioned. For procedures that allow the 
simultaneous treatment of space-time systems, see Myers (1992) and 
Panesar (1998). A short and incomplete list of available software is given 
at the end of this section. 

Potential use of spatial statistics 

Geostatistics provides useful tools for spatial analyses, creation of optimized 
maps and simulation of spatial processes. For example, geostatistics can 
be used to obtain estimates of the total nitrogen accumulation in the soil 
under a legume fallow, or total phosphorus availability in the soil of an 
agroforestry plot with complex spatial patterns of fertilizer application and 
nutrient uptake by plants. When creating maps of chemical or physical 
properties of a field or landscape, information on the similarity of values 
from neighbouring samples and increasing dissimilarity of values with 

R. Coe ef al. 

increasing distance between the samples, the so-called autocorrelation 
structure for the area, can be used to calculate weighting factors which 
minimize the estimation error of the variables under study (Goovaerts, 
1998). Geostatistical methods can also be used for the generation of 
optimized sampling schemes according to the requirements of the 
experiment, financial constraints or predefined local accuracy (Webster 
and Burgess, 1984), thereby reducing the experimental effort. Further, it 
is possible to identify optimal pixel sizes for process models (Wade et al., 
1996) or to generate parameter fields for the analysis of the transport of 
nutrients or pollutants in structured soils (Piehler and Huwe, 2000). Lopez 
and Arrue (1995) used geostatistics to optimize an incomplete block design 
for tillage experiments which proved to be considerably more efficient 
than the corresponding complete block design. 

Although geostatistical methods have been widely used in the field of 
soil, agricultural and hydrological sciences at different scales (Oliver, 1992; 
Schiffer, 1992; Webster and Boag, 1992; Huwe, 1993; Hoosbeck and 
Bouma, 1998), not much work on spatial analysis has been reported so far 
in the field of agroforestry. However, potential benefits for agroforestry 
studies are demonstrated by applications in forest and agricultural 
ecosystems (Sylla et al., 1996; Bragato and Primavera, 1998; Gorres et al, 
1998), orchards (Gottwald et al. , 1995), shrub-steppe ecosystems (Smith et 
al, 1994; Halvorson et al., 1995) and in soil surveys (Di et al, 1989). The 
scale of application ranges from the microscale of soil pores (Grevers and 
Jong, 1994) through agricultural fields (Bragato and Primavera, 1998) and 
landscapes (Wade et al., 1996) to regions of several thousand square 
kilometres (van Meirvenne et al., 1996). Processes involved in the studies 
comprise tree growth (Meredieu et al, 1996), crop yield (Wopereis et al, 
1996), water transport in soils (Mulla, 1988), evaporation (Lascano and 
Hatfield, 1992), carbon and nitrogen mineralization (Smith et al., 1994), 
nitrous oxide fluxes (Velthof et al., 1996) and pesticide transport to the 
groundwater (Soutter and Pannatier, 1996). As water and matter fluxes 
play an increasing role as indicators for sustainability and environmental 
compatibility, the aforementioned aspects of heterogeneity will be of 
increasing importance in future agroforestry studies. The epidemiology 
of plant diseases and even the spatial distribution of leaves and fruits in 
trees have also been studied by means of geostatistical methods (Chellerni 
et al, 1988; Monestiez et al, 1990; Orum et al, 1999). 

General principles of geostatistics 

Spatial variability, or spatial patterns, can conceptually be divided into 
components which are caused by either deterministic or stochastic 
processes. Apparently stochastic behaviour of a system may be induced by 

Designing Experiments and Analysing Data 69 

so-called hidden variables (i.e. variables that were not measured), or simply 
by the scale of measurements. For example, to predict nitrate leaching in 
the soil, we may use an average root length density and average hydraulic 
conductivity per soil layer as determined from a number of field 
measurements as input variables into a deterministic model. However, 
even in the best of cases the output of this model will not be correct for 
every single spot in the area, because not every spot has average properties 
with respect to root length density or hydraulic conductivity and, of course, 
we can never know the exact root length density and hydraulic 
conductivity at every spot in the soil. Thus, whereas part of the factors that 
influence nitrate fluxes are 'known' (i.e. have been described through 
measured, average values), another part will always remain unknown and 
will introduce unexplained (apparently stochastic) variability in the 
behaviour of the system. As mentioned before, this unexplained variability 
can be so large in certain cases that the model result will be completely 
wrong, simply because the behaviour of the system is not determined by 
average but rather by extreme values. 

The identification, analysis and description of the deterministic system 
components belong to the field of deterministic, process-based modelling, 
whereas the analysis and modelling of stochastic patterns is done within 
the framework of the theory of stochastic processes, including geostatistical 
theory. Classical statistical methods which are generally used in 
agroforestry largely ignore spatial structures. Most tests assume 
independence of sampling points and, in addition, normally distributed 
variables or residuals. By ignoring spatial autocorrelation structures, an 
error is introduced into the analysis. 

The fundamental backbone of geostatistics is the concept of 
regionalized variables, which is based on the work of engineers, 
mathematicians, physicists and biometricians in the early 1940s (Mathes 
and Ries, 1995). The theory in its present form was developed mainly by 
Matheron (1963). For original literature and more complete descriptions 
of geostatistical theory and applications see Journel and Huijbregts (1978), 
Journel (1989), Bardossy (1992) and Webster and Oliver (2001). 

The main idea behind geostatistics is the common observation in the 
field that values of variables often resemble each other more as the distance 
between sampling points decreases. With increasing distance, the spatial 
influence of neighbouring samples becomes smaller, and above a certain 
limit, the so-called range, variables are independent in a statistical sense. 
Measurements are treated as realization of an ergodic spatial stochastic 
process, which means that the characteristics of the stochastic pattern or 
process can be derived from this single spatial data set. For example, a 
single measurement of soil phosphorus in several sampling positions in a 
plot is enough to describe the stochastic pattern of phosphorus distribution 
in the plot. Another assumption of geostatistical theory is intrinsic (or weak) 


R. Coe ef al. 

stationarity, which means that for any two sampling points in the area, the 
difference in the measured value (e.g. soil phosphorus content) depends 
only on the distance between the two points (the lag), independently of 
where in the plot the two samples were collected. This assumption is often 
problematic, especially in areas with pronounced soil gradients or abrupt 
changes in soil properties, as will be discussed further below. 

Procedure of geostatistical analysis 

A basic component of geostatistical analysis is the semivariogram (Fig. 3.7). 
A semivariogram is a mathematical function that describes how two 
sampling points become more different with increasing spatial distance 
between them, until a certain distance when the points become 
independent from each other. The semivariogram describes the auto- 
correlation structure of the measured variable in the respective area. The 
mathematical formula of the semivariogram is given further below. 
Geostatistical studies involve the following steps (Kitanidis, 1997): 

• design of the sampling grid; 

• data collection; 

• analysis of spatial structure (determination of empirical semivario- 






j^~ — T - 

' "N 

V sill 


r nugget variance 

I I I 

l l 




Fig. 3.7. Schematic semivariogram with a nugget variance of 2, a sill of 7 and a 
range of 7. 

Designing Experiments and Analysing Data 71 

• selection of an appropriate semivariogram model; 

• interpolating the grid by best, linear, unbiased estimations (BLUES); 

• interactive optimization of the semivariogram model using cross- 
validation techniques. 

The design of the sampling grid (dimension, grid size, regular or 
irregular spacing) depends on problem-specific accuracy requirements 
and the spatial structure of the sampling area. One-dimensional grids 
(transects) are used for linear structures, such as gradients between linear 
tree plantings and adjacent fields. Two- and three-dimensional grids are 
used for maps, for example of nutrient distribution in a plot, or for the 
determination of parameter fields for process simulations, such as the 
transport of nutrients and pollutants in soil. Webster (1985) found that for 
isotropic semivariograms, i.e. in areas where soil properties are expected 
to change according to the same pattern in all directions, the best two- 
dimensional sampling scheme is a regular, equilateral triangular grid. For 
the estimation of mean values of square blocks, the best position of 
sampling points is in the centre of the blocks (Webster and Burgess, 1984). 
However, in practical studies it is often not possible to assure a regular 
grid. Further, in many cases it is advantageous to use a nested grid in order 
to determine the semivariogram for short sampling distances (Selles et al., 
1999) and to focus the accuracy of the calculated map on a predefined area 
of special interest (e.g. a sampling plot within a larger section of the 

A qualitative explanation of the semivariogram has been given above. 
The mathematical formula of the semivariogram y(h ) is given by 

where z(x ) denotes the measurement at location x , and N is the number 
of pairs with a distance of h . The arrow indicates vectors. In simple words, 
the semivariogram gives for every distance h between two data points a 
mean squared difference, obtained by averaging all pairs of data points 
with this distance in the data set. For a small h (samples collected close to 
each other), this difference will be relatively small, and for a larger h it will 
be greater. The use of vectors in the formula indicates that the 
semivariogram may not be the same in all directions, i.e. it may be 

As illustrated in Fig. 3.7, typical semivariograms are characterized by: 
(i) a short-range variability, the nugget variance, which is determined by 
the closest sampling distance in the grid and is therefore to a certain extent 
an artefact; (ii) an asymptote equal to the background variance of the 

72 R. Coeefa/. 

variable, i.e. the variance between sampling points that are completely 
independent of each other at the scale of the study; and (iii) a maximum lag 
value that characterizes the range of spatial influence and is therefore called 
range. Points that are farther apart than this range are independent at the 
scale of the study. The sill is the difference between the background variance 
and the nugget variance. For irregular grids and anisotropic conditions, the 
differences are grouped into several distance and angle classes. 

The experimental semivariogram is the basis for the selection of a 
theoretical model, which is then used for the analysis of spatial structures 
of the variable under study. Theoretical semivariogram models are 
mathematical functions that resemble the empirical semivariogram as 
closely as possible. Not all functions that would fit the data can be used as 
semivariogram models. If the model is to be used for spatial estimates or 
spatial simulations, it must be positively defined, a mathematical 
requirement that will not be discussed here (see Journel and Huijbregts, 
1978). Unfortunately, it is not trivial to prove whether a given function 
fulfils this requirement or not. Custom models that are known to fulfil this 
criterion are the linear, spherical, exponential, logarithmic and Gaussian 
models. Other positively defined models are the quadratic, rational 
quadratic, power and wave models (Alfaro, 1980; Cressie, 1991; Pannatier, 
1996; Kitanidis, 1997). Models for anisotropic conditions are also available. 
The field of variography is still in development. 

A first estimate of semivariogram model parameters can be obtained 
by a least squares approach using the empirical semivariogram data. In a 
subsequent procedure, a number of different kriging techniques (see 
below) can be used to optimize the semivariogram and yield the best- 
possible estimated data. A criterion for the quality of the estimation is 
obtained by jack-knifing, a cross-validation procedure where each 
measured point is estimated from the neighbouring points and the 
estimated values are compared with the true values. The correct 
optimization algorithm for the semivariogram thus includes the whole 
estimation procedure, including the krige algorithm itself (Kitanidis, 1997), 
and not only the fitting of measured and estimated semivariogram data. 

Point kriging 

Kriging is a linear unbiased estimator for the variable under study. It is 
calculated as the weighted mean of the values of all sampling points or of 
the points in a user-defined local neighbourhood (which often ascertains 
local stationarity and thus avoids trend corrections, see below): 

= 1* 


Designing Experiments and Analysing Data 73 

where z. denotes the measured values, z the estimate at a user-defined 
location and X,the weighting factors calculated from the krige system. 
Kriging is also BLUES - the best, linear, unbiased estimator. Kriging, by 
definition, minimizes the estimation variance, which is the variance 
between the measured and the estimated data points. For each estimation 
on a location between the grid nodes, kriging involves the solution of a 
linear equation system, the krige system, that yields the weighting factors 
X t for the linear estimation procedure and a so-called Lagrange multiplier ll, 

which results from the consideration of the constraint Y^ = 1 . With the Xs 


and (J it is possible to calculate the estimation variance for each estimated 
value and thus to determine confidence intervals. If the stochastic 
assumptions of the krige procedure are fulfilled, we may thus obtain not 
only maps with estimated (rather than mean) values, but in more or less 
the same step also maps that show the reliability of the estimated data, a 
source of information that is not provided by any other interpolation 

Special kriging techniques 

Point kriging, as described above, is the simplest kriging technique and 
for many applications in agricultural or environmental sciences it is 
sufficient. For example, maps of soil texture or pH can be generated with 
this technique. There are also several advanced kriging techniques for 
special data structures, quality aspects and project goals that are briefly 
introduced below. 


A serious violation of the intrinsic model assumption of weak stationarity 
is given by a spatially non-constant expectation of the random variable 
under study, i.e. a trend. Trends may be caused by gradients in soil 
conditions in sloping areas, gradually changing geological or climatic site 
conditions, or land-use history. Common techniques to avoid the 
associated errors are the moving neighbourhood algorithm and the 
universal kriging approach. While the moving neighbourhood technique 
assumes local stationarity and thus restricts the grid points used for 
estimation to this neighbourhood, universal kriging assumes a predefined 
functional behaviour of the trend and eliminates this trend globally in an 
iterative algorithm (De Marsily, 1986; Deutsch and Journel, 1992). These 
techniques are available in some of the software packages listed at the end 
of this section. 

74 R. Coeefa/. 

Block kriging 

Sometimes it may be desirable to estimate not point values but values that 
represent the average of the variable in a defined volume around the 
estimation point. Examples include the estimation of nutrient stocks or 
plant-available water in a given soil volume. The average variable is 
obtained by integrating over a volume V: 


dv (3.4) 

The estimation process is analogous to point kriging and differs mainly 
in the averaging of the semivariances in the system of linear equations. As 
a consequence of averaging, the estimation variances are typically much 
smaller than those obtained by point kriging (De Marsily, 1986; Deutsch 
and Journel, 1992). Gorres et al. (1998) used block kriging for the 
generation of maps of carbon mineralization, bulk densities and nematode 
densities in a forest and agricultural field. 


In most field studies, a number of parameters are measured at each 
sampling point rather than a single value. Many of these parameters may 
be intercorrelated as well as autocorrelated. Thus, taking into account the 
complete correlation structure should yield better estimation results 
compared to separate kriging of every single variable. Co-kriging was used 
by Halvorson et al. (1995) to analyse the spatial relations between resource 
islands in the soil and different plant species in a shrub-steppe ecosystem, 
and similar applications in savannas and parkland systems could be useful. 
As in simple kriging, co-kriging minimizes the estimation variance and 
provides its value for each location. Extensions to more than two correlated 
variables are possible (De Marsily, 1986). 

Indicator kriging 

This is a technique that may become particularly useful in environmental 
decision problems, such as when an environmental measure depends on 
the percentage of values above or below a given threshold in the model 
area (Akin and Siemes, 1988; Mulla and McBratney, 1999). Indicator 
kriging requires the transformation of the variables into indicator variables 
(values above a defined threshold are assigned a value of 1, values below 
the threshold are assigned a value of 0). Variography and kriging are 
conducted as described above. Kriging is carried out as simple point 

Designing Experiments and Analysing Data 75 

kriging. Spatial averaging of the kriged indicator values yield the 
proportion of values of the original data set that are below the threshold. 
Halvorson et al. (1995) used this technique in the aforementioned study 
to determine the probability that certain combinations of variables 
occurred at unsampled locations. 

Ceostatistical simulation 

Kriging is an interpolation algorithm that is optimal in the aforementioned 
sense. However, it generates more or less smooth maps that are in some 
ways unrealistic when compared with the real situation in the field. 
Unfortunately, this not only is of interest from an aesthetic point of view 
but may even yield incorrect results. This is always the case when extreme 
values are of importance. For example, transport of nitrate or pesticides 
in soil is often governed by preferential pathways, caused by macropore 
systems or spatial heterogeneity of soil hydraulic properties. Using simple 
similarity concepts together with geostatistical simulation, it is possible to 
generate heterogeneous parameter fields. Effects of this simulated 
heterogeneity for transport processes may then be calculated with 
numerical transport models (Piehler and Huwe, 2000). For example, a 
critical parameter of denitrification in soils is the soil water content. As 
kriging is a smoothing estimator, the percentage of water content events 
above a critical value, and consequently denitrification and the production 
of nitrous oxides, may be underestimated. Conditional geostatistical 
simulation provides a measure to generate more realistic maps that are 
compatible with the determined semivariogram. Furthermore, in the grid 
nodes the simulated values are identical to the measured values. Several 
algorithms for spatial simulation are available (e.g. the random coin 
method by Alfaro, 1980). A comprehensive overview with algorithms and 
FORTRAN -codes is given in Deutsch and Journel (1992). 

Alternative methods and tools 

Geostatistical methods have some serious drawbacks. Although normal 
distribution of the random variable is not explicitly required, it is 
recommended by some authors and may facilitate the interpretation of 
the maps of estimation variances. The applicability of geostatistics is 
restricted to situations where ergodicity may be assumed and the intrinsic 
hypothesis is valid. Thus, trends and structural discontinuities render the 
geostatistical analysis more difficult. Such situations may be caused by 
short-range, sharp gradients in a structured agroforestry system as well as 
by marked changes across a landscape (e.g. the boundary between 

76 R. Coeefa/. 

grassland and forest), and geostatistics may then not be the appropriate 

Recent developments in mathematics and statistics provide algorithms 
and tools that also have potential for spatial statistics, although up to now 
they have not been studied and tested in full detail. Most of the methods 
have in common that they are robust and do not prescribe probability 
density functions like the bell-shaped Gaussian distribution. See Mathes 
and Ries (1995) for local gradient analysis, Batchelor (1998), Levine et al. 
(1996) and Liescheid et al. (1998) for artificial neural networks, and Woldt 
et al. (1992) for a geostatistical application of fuzzy set theory. 

Ceostatistical software 

Some information is given below about software available in the public 
domain ( PD ) and commercially (®). 

• Geo-EAS pd and GeoPack™ are DOS-programs that provide basic 
functionality of variogram analysis, kriging and visualization ('rough 
and dirty'). The DOS-environment is not very user friendly and does 
not fit into modern Windows systems ( 

• surfer® (Golden Software, 1999) is a contouring and surface mapping 
package that provides variogram analysis, variogram fitting and 
kriging, including block kriging and universal kriging 

• gslib™ (Deutsch and Journel, 1992) is a most powerful collection of 
FORTRAN -programs that covers almost every aspect of geostatistics. 
The handbook is a very good overview of geostatistical theory. The 
user must be familiar with FORTRAN and have access to a FORTRAN 

• systat® is a commercial statistical package that now includes 
geostatistical routines based on the gslib package and CART- 
Algorithms ( science/SYSTAT). 

• os+®is a commercial geostatistical software package with a user 
friendly interface ( 

• Data-Engine® provides algorithms for fuzzy control, artificial neural 
networks and combinations of both ( 

• Stuttgart Neural Network Simulator (SNNSPD) is a very powerful 
artificial neural network ( 

• VarioWin (Pannatier, 1996) is an easy to use software package for 
variogram analysis and visualization. 

Chapter 4 

Soil Organic Matter 

G. Schroth, 1 B. Vanlauwe 2 and J. Lehmann 3 

1 Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (INPA), CP 478, 6091 1-970 Manaus, 
AM, Brazil; 2 Tropical Soil Biology and Fertility Programme, PO Box 
30592, Nairobi, Kenya; 3 College of Agriculture and Life Sciences, 
Department of Crop and Soil Sciences, Cornell University, 909 
Bradfield Hall, Ithaca, NY 14853, USA 

4.1 Synopsis 

Soils contain a variety of organic materials, ranging from living roots, fauna 
and microbes, through dead tissues in various stages of decomposition, to 
relatively stable, dark-coloured transformation products with no 
discernible anatomical structures, the so-called humus (Jenkinson, 1988b). 
Following Jenkinson (1988b), the term soil organic matter includes all 
organic substances in soil, both living and dead, although the living 
organisms (the soil biomass) usually contribute less than 5% to the total 
soil organic matter. Certain terminologies do not include the biomass in 
soil organic matter, and some authors use the terms humus and soil 
organic matter synonymously, so care is necessary to avoid mis- 
understanding. In practical terms, analyses of soil organic matter are 
usually carried out on air-dried soil that was passed through a 2 mm sieve 
and will include all living and dead organic materials in the soil that are 
not removed by this procedure (Anderson and Ingram, 1993). In studies 
relating soil organic matter contents or fractions to soil aggregation, 
particles >2 mm may also be included in the analysis. The organic 
materials that lie on the soil surface, including naturally fallen litter, crop 
residues, mulch materials and their decomposition products, are normally 
sampled and analysed separately from the soil (see Chapter 6). Because 
decomposing roots have a function similar to that of above-ground litter 
in the replenishment of soil organic matter, they are sometimes called 
below-ground (or root) litter. This should not be confused with the term 
soil litter, which has been used for a density/particle size fraction of soil 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 77 

F.L. Sinclair) 

78 C. Schroth ef al. 

organic matter (see Section 4.3). A detailed discussion of the chemistry, 
formation and transformations of soil organic matter is provided by 
Stevenson and Cole (1999). 

The organic carbon contents of terrestrial soils vary between < 1% in 
many sandy soils and about 3.5% in grassland soils. Poorly drained soils 
may have much higher organic carbon contents (Stevenson and Cole, 
1999). Contrary to widespread belief, the organic matter contents of 
tropical soils are not generally lower than those of temperate soils; in fact, 
they vary with vegetation (higher in forest than in savanna soils), climate 
(higher in mountain forests than lowland forests), soil texture (increasing 
with increasing clay and silt content of the soil), mineralogy (higher in 
volcanic soils due to the stabilizing effect of allophane on soil organic 
matter) and soil use (Feller et al., 1991b; van Wambeke, 1992). 

Role of soil organic matter in soil fertility 

The organic matter content of a soil influences a wide range of soil 
properties and processes and, despite the fact that plants do not require 
organic matter as such for their growth and development, it is considered 
one of the most important components of soil fertility (Gregorich et al., 
1994). This is especially so in agriculture with low external inputs. In soils 
dominated by low-activity clays such as kaolinite, the cation exchange 
capacity, and thus the ability of the soil to retain nutrients against leaching 
in a plant-available form, depends strongly on its organic matter content. 
Soil organic matter may act either as a source or a temporary sink of 
nutrients such as nitrogen, phosphorus and sulphur (see Sections 5.2- 
5.4). Organic substances can increase the solubility of phosphorus in the 
soil and reduce phosphorus fixation by competing with phosphate ions 
for sorption sites such as oxides of iron and aluminium and clay minerals 
(especially allophane in volcanic soils). They can also complex aluminium, 
iron and calcium cations, which would otherwise form precipitates of low 
solubility with phosphate. Complexation by organic matter also detoxifies 
soluble aluminium in acid soils (Stevenson and Cole, 1999; Palm et al., 

Whereas the regulatory functions of soil organic matter on nutrient 
availability and acidity are most important in land-use systems with low 
addition rates of mineral fertilizers and lime, the stabilization of soil 
structure by organic matter is equally important in low as in high input 
agriculture. This is especially so in sandy soils where the stabilizing effect 
of clay on soil structure is limited by low clay contents (Pieri, 1989). In 
contrast, Ferralsols have an inherently favourable microaggregate 
structure which is little influenced by their organic matter content (Dudal 
and Deckers, 1993; Lehmann et al, 2001a). The organic matter content 

Soil Organic Matter 79 

of a soil also influences its ability to store water, although this effect should 
not be overestimated (Gregory, 1988; de Ridder and van Keulen, 1990). 
For example, Diels et al. (2002) estimated that, in a soil with sandy loam 
texture, an increase in carbon content from 0.8 to 1.3% in the top 15 cm 
of the profile allowed an additional water storage of only 1 mm, compared 
with at least 50-70 mm of available water in the root zone of a 4-week-old 
maize crop. 

Soil organic matter is also a substrate for soil biota, and it can help to 
control populations of plant pathogens in the soil by stimulating the activity 
of antagonistic microorganisms (Heitefuss, 1987). Recently, the potential 
role of soil organic matter as a sink for atmospheric carbon dioxide has 
received attention (van Noordwijk et al., 1997). 

Effects of land use on soil organic matter 

Because of the importance of organic matter for soil fertility, the effects of 
land-use change, management practices and fallowing on the organic 
matter content of tropical soils have often been investigated. Numerous 
studies in the tropics have demonstrated loss of soil organic matter and 
organic nutrients following the conversion of forest or savanna ecosystems 
into either annual (Nye and Greenland, 1960; Pieri, 1989; de Ridder and 
van Keulen, 1990; du Preez and du Toit, 1995) or perennial cropping 
systems (Ollagnier et al, 1978; Ahenkorah et al, 1987; Ekanade, 1987). 
The reasons for this include a reduction in total organic inputs (litter, crop 
residues, manure) and increased mineralization rates of soil organic matter 
caused by tillage, increased soil temperatures due to exposure of the soil 
surface, and increased wetting-and-drying cycles. Organic matter may also 
be lost by soil erosion (Pieri, 1989; Palm et al., 2001). Fallowing leads 
eventually to increases in soil organic matter content, reversing the effects 
of land clearing and cropping, although decreases may continue to be 
observed during the first 1 or 2 years of a new fallow (Szott et al, 1999). 
In situations where soil organic matter content is either increasing or 
decreasing, new equilibria are reached more rapidly in sandy than in clay 
soils (Feller et al, 1991b; du Preez and du Toit, 1995). Like fallows, trees 
in crop fields can contribute to the maintenance and improvement of soil 
organic matter levels through increased inputs of litter and roots, 
reduction of soil temperatures through shading, and soil protection from 
erosion. Favourable effects of agroforestry on the soil organic matter 
balance have been demonstrated in many studies (Young, 1997; Rao et al. , 

Several management factors influence the rate of organic matter loss 
under cultivation in a manner that is often complex and difficult to predict. 
Additions of organic materials such as root biomass, manure and compost 

80 C. Schroth ef al. 

have a favourable effect on soil organic matter (de Ridder and van Keulen, 
1990). In contrast, the incorporation of cereal straw has been shown to 
stimulate microbial activity and cause carbon losses in sandy savanna soils 
in West Africa (Pieri, 1989). Measures that increase the amount of plant 
biomass above and below ground and its rate of accumulation tend to 
improve the organic matter balance by increasing carbon inputs to the soil 
and may also reduce erosion of otherwise incompletely covered soils, 
especially during the early part of the cropping season (Pieri, 1989). Most 
of the available data from trials in the West African moist savanna zone 
indicate that organic carbon contents of plots with fertilizer application are 
comparable to or slightly higher than those of unfertilized plots (Vanlauwe 
et al., 2001). Sanchez (1976) also reports that long-term application of 
inorganic fertilizer delays the decrease in soil organic matter content under 
cropping by providing more crop residues, including roots. However, in 
contrast to the generally favourable effect of an equilibrated fertilization 
on soil organic matter, large applications of nitrogen fertilizer such as those 
recommended under intensive production of cereals have been shown to 
increase soil organic matter loss in the West African savanna (Pieri, 1989). 
For maintaining adequate levels of organic matter in the soil, sufficient 
amounts of organic inputs are needed, in the form of either crop residues, 
tree litter, compost, manure or a combination of these, depending on the 
land-use system. Long-term experiments on a range of soils in both West 
Africa and India show declining yields of a range of crops when only 
inorganic fertilizer is used for prolonged periods, suggesting that organic 
matter generally declines to below a critical level under continuous 
cropping unless there are organic additions (Greenland, 1994). 

The effect of tillage on the soil organic matter balance is also variable, 
as tillage may increase crop yields and therefore carbon inputs into the 
soil, but also increases the microbial decomposition of soil organic matter 
and in some cases soil erosion (Pieri, 1989). 

Functional pools of soil organic matter 

As mentioned initially, soil organic matter is a mixture of living and dead 
components differing widely in their chemical composition and physical 
structure, which make them more or less resistant to microbial degradation. 
Some components are characterized by relatively rapid turnover and may 
contribute significantly to the mineralization and immobilization of 
nutrients. Examples are the microbial biomass and plant residues which 
have not yet undergone an intensive transformation in the soil (the light 
fraction or particulate organic matter fraction, see below). These fractions 
also respond rapidly to management measures. Humic substances, in 
contrast, are protected to some extent from microbial degradation by their 

Soil Organic Matter 

chemical structure, association with mineral particles and/or occlusion into 
aggregates. These substances make an important contribution to soil 
fertility through the stabilization of the soil structure and cation exchange 
processes, whereas their contribution to nutrient release is insufficiently 
known (Woomer et al. , 1994). Recently, increased attention has been given 
to finely divided charcoal as a component of the stable organic matter of 
many soils (Haumaier and Zech, 1995; Skjemstad et al., 1996). 

Considerable research effort is being invested in the identification and 
characterization of different pools of soil organic matter. The objective of 
these studies is to obtain an improved understanding of the 
transformations, stabilization and loss of soil organic matter, and of the 
relationships between the characteristics of certain organic matter fractions 
and their functions in soil, especially their nutrient supply properties. 
These attempts are associated with the development of computer 
simulation models, which divide soil organic matter into several pools with 
different turnover times to predict effects of site factors and management 
measures on carbon stocks, nutrient cycling and plant growth (see Box 
4. 1). For agroforestry, these efforts are relevant in so far as they may lead 
to prediction of the effects of certain agroforestry techniques, plant species 
and management measures on the soil carbon balance and associated 
processes, such as nutrient release and stabilization of soil structure. 

Another application of soil organic matter fractionation is that, by 
quantifying those carbon fractions that respond rapidly to changes in 
vegetation cover or soil management, the effect of land-use change or 
experimental treatments on soil organic matter can often be detected more 
sensitively over the course of an experiment than when only total soil 
organic matter is measured. 

Box 4.1. Modelling soil organic matter dynamics. 

The amount of organic matter in a soil is a function of additions of fresh organic 
matter and decomposition rates, the latter being influenced by a series of intrinsic 
(e.g. residue quality, see Section 6.4) and soil-related controls (e.g. soil water 
content and temperature, composition and activity of the decomposer community 
and soil texture). Models describing soil organic matter dynamics usually contain 
a number of conceptually distinct pools, interconnected through a number of mass 
fluxes. A first set of pools is usually related to fresh organic matter and takes 
account of its quality, whereby higher quality is related to faster decomposition. 
Paustian etal. (1 997) present a concise summary of mathematical approaches used 
to include residue quality in decomposition models. A second set of pools 
differentiates various soil organic matter fractions with varying turnover times. The 
microbial biomass, which is the soil organic matter pool with the highest turnover 
and through which most of the added organic carbon passes, is often one of these 


82 C. Schroth ef al. 

Box 4.1 . Continued. 

fractions. Besides carbon, certain models such as century (Parton et al., 1994) 

also simulate nitrogen and phosphorus dynamics. 

The level of complexity of soil organic matter models increases with the 
number of pools and fluxes. Simple models usually contain a substantial amount 
of empirical information, obtained through calibration against available 
measurements for a certain environment. The validity of these parameters is 
obviously limited to that environment. More complex models contain a more 
important range of differential equations mechanistically linking various soil 
organic matter pools. A major advantage of simple models containing a minimal 
amount of pools and fluxes is that such models can be analytically solved and do 
not require simulation techniques. An example of a simple model which can be 
analytically solved is the icbm model, containing only two pools, two decay 
constants, and parameters for litter input, humification and external influences 
(Andren and Katterer, 1997). An example of a model aiming at the mechanistic 
understanding of decomposition is the foodweb model of Hassink et al. (1994). 
As more complex models require more parameters to be calibrated, it is advisable 
to keep models as simple as possible if the modelling exercise is not the purpose 
of the study in itself. 

Several reasons can be identified for embarking on soil organic matter 
simulation studies. First, as mentioned earlier, soil organic matter consists of a 
whole range of organic materials with varying resistance to microbial degradation 
and is continuously replenished by various sources of fresh organic matter. 
Summarizing available knowledge in the form of a simulation model can 
substantially help to understand how the different organic materials and soil 
organic matter pools interact and to develop research hypotheses. Secondly, after 
changes in land use, it can take several years before soil organic carbon reaches 
a new steady state. Although trials quantifying these changes in soil carbon are 
needed to calibrate the long-term modules of any soil organic matter model, it is 
clear that trials cannot be set up to deal with every possible land-use scenario. A 
well-calibrated model can be used to evaluate the impact of various land-use 
scenarios on the final soil organic carbon content and guide scientists and policy 
makers to develop guidelines favouring carbon build-up. Thirdly, models that 
quantitatively account for soil- and environment-related modifiers of the 
decomposition process allow extrapolation of carbon dynamics from the field scale 
to a wider region and assist in developing regional carbon profiles. 

Several attempts have been made to link conceptual soil organic matter pools 
with experimentally determined soil organic matter fractions. The only soil organic 
matter fraction that has been rather successfully linked to experimental data is the 
soil microbial biomass. Other clear links between experimentally determined soil 
organic matter fractions and conceptual pools are rare, although the light 
particulate organic matter fraction (>100 urn, <1 .4 g crrr 3 , see Section 4.3) has 
been equated with the Added Organic Matter pool in the daisy soil organic matter 
submodel (Mueller era/., 1998). Balesdent (1996) successfully related physical 
separates of soil organic matter with the structural compartment of the Rothamsted 
carbon model, and Gaunt et al. (2000) proposed a physical soil fractionation 
scheme to obtain organic matter fractions suitable for modelling. 

Soil Organic Matter 83 

Most soil organic matter models have been developed based on information 
obtained under temperate conditions. Although it is often argued that processes 
are not fundamentally different between temperate and tropical regions (Jenkinson 
and Ayanaba, 1977), environmental conditions, soils and vegetation are so 
different that models developed under temperate conditions cannot be directly 
applied to tropical conditions without proper validation. Examples of models 
which have been successfully used to simulate carbon behaviour under tropical 
conditions are the century model (Parton etal., 1 994) and the Rothamsted carbon 
model (rothic) (Jenkinson, 1990; Diels etal., 2002). 

Labile soil organic matter 

Particular attention has been given to microbial biomass (see Section 4.5) 
and to the light fraction and particulate organic matter fraction, which are 
obtained from soil samples by either density separation or sieving or a 
combination of both (see Section 4.3). These carbon pools are 
characterized by relatively rapid turnover in the soil and have, therefore, 
been used to estimate an active, labile, nutrient-supplying fraction of soil 
organic matter (Palm et al., 2001). 

The microbial biomass represents only a few per cent of soil organic 
matter, but is essential for decomposition and nutrient release from 
organic materials and contributes to soil aggregation. Microbial biomass 
carbon can be used as an index of soil organic matter dynamics by relating 
it to total soil carbon and comparing with a local reference soil under 
natural vegetation (Gregorich et al., 1994). 

The light fraction and particulate organic matter consist mainly of 
plant residues, principally roots, with some residues of animals and 
microorganisms. The turnover time of these fractions ranges from a few 
days to a few years (Woomer et al, 1994; Stevenson and Cole, 1999). Seeds 
and charcoal may also be present and may have to be removed as they do 
not belong to the labile organic matter (Barrios et al. , 1997). Although the 
light fraction contains most of the particulate organic matter, the two 
fractions are not identical in quantity and composition (Gregorich et al., 
1994). The light fraction may contain as much as 30-40% of the organic 
carbon of a soil (Gregorich and Ellert, 1993; Stevenson and Cole, 1999), 
but values of 10% or less have been reported for sandy savanna soils in 
Africa (Barrios et al., 1997; Lehmann et al., 1998b). An important part of 
the soil microbial and enzyme activity is associated with these fractions. 

Light fraction and particulate organic matter respond rapidly to 
changes in soil management and are affected by litter inputs and 
decomposition conditions in the soil. The greatest amounts in the soil are 
measured immediately after the incorporation of plant residues. 
Decomposition in the soil progressively removes the most active, nutrient- 

84 C. Schroth ef al. 

supplying components, so that later measurements include increasingly 
recalcitrant materials with low nutrient-supply properties (Palm et al., 
2001). These seasonal fluctuations in quantity and quality of the light or 
particulate organic matter fraction have to be taken into consideration 
when developing sampling plans and interpreting the results. In fact, a 
recent study reported increased particulate organic matter in a Ferralsol 
under a tree crop with slowly decomposing litter compared with trees that 
produced litter of higher quality (Lehmann et al., 2001a), suggesting that 
this fraction contained mostly recalcitrant residues. 

An alternative approach to the separation of a labile soil organic matter 
fraction which could help to circumvent some of the difficulties associated 
with the heterogeneous nature of physical fractions is selective oxidation 
with chemicals. Blair et al. (1995) used permanganate to separate the labile 
from the stable soil organic matter pool. Based on this two-pool 
fractionation, they proposed a carbon management index which is 
obtained by multiplying the carbon pool index (quotient of total carbon 
in the sample soil and in a reference soil) by the lability index (quotient of 
the labile and the stable fractions in the sample soil). The carbon 
management index was sensitive to different management regimes in a 
number of experiments (Blair et al., 1997). Systematic comparisons of 
physical and chemical indices of labile soil organic matter pools for 
different pedoclimatic and management conditions are clearly needed. 
Combinations of both approaches are possible. 

Formation of stable soil organic matter 

As mentioned initially, the supply of nutrients is only one of several 
important functions of soil organic matter, others include the stabilization 
of soil structure, water-holding capacity and cation exchange properties. 
Therefore, the build-up of a labile, nutrient-supplying, soil organic matter 
pool, which is rapidly mineralized in the soil, is not the only objective of 
organic matter management in agroforestry, another objective of similar 
importance is the maintenance of a high content of stable organic matter. 
Therefore, a question of critical interest is how the build-up of stable 
carbon in the soil may be enhanced by, for example, the use of certain 
plant species in fallows or tree-crop associations, the treatment of biomass 
before it is applied to the soil (composting), and the method and timing 
of biomass application. 

Unfortunately, the processes of the stabilization of organic matter in 
soil are still poorly understood, and it is not known which characteristics 
of an organic material determine whether it enters a labile fraction of soil 
organic matter and is rapidly mineralized, or is humified and ends up in 
a stable pool. Some studies have indicated that biomass with a higher C:N 

Soil Organic Matter 85 

ratio and lignin content results in more soil organic matter, although this 
has not always been confirmed (Palm et al, 1997). In a decomposition study 
with biomass from three tree species in a sandy savanna soil in Togo, the 
biomass with the highest polyphenols ratio and the slowest 
decomposition (from Senna siamea) was the most efficient in increasing the 
carbon and nitrogen content of the silt and clay fractions of the soil 
(Lehmann et al, 1998b). This concurs with the suggestion that materials 
rich in reactive polyphenols could increase soil organic matter formation 
(Palm et al, 2001). In a study in Nigeria, slowly decomposing organic 
inputs with high C:N and lignin:N ratios were also shown to produce 
organic matter with a higher cation exchange capacity than other 
materials, suggesting that they could be useful for increasing the cation 
exchange capacity of highly weathered soils (Oorts et al, 2000). 

It is, however, likely that materials which are most efficient for soil 
organic matter formation such as those high in lignin and polyphenols, 
are not the same as those that release nutrients immediately and increase 
crop yields in the short term by decomposing quickly because they have a 
low (lignin + polyphenol): N ratio (see Section 5.2 and Chapter 6). Faced 
with this choice, farmers are likely to select the organic materials (or plant 
species) that offer the greater short-term benefits, rather than those that 
offer the greatest effect on the long-term soil carbon and nutrient balance. 
Biomass rich in both lignin and nitrogen could possibly combine soil 
organic matter formation with high nitrogen availability, but this remains 
to be confirmed by further research (Palm et al., 2001). The outcome of 
such research would determine whether mixtures of tree species with 
contrasting properties, such as fast and slowly decomposing biomass, are 
desirable in agroforestry designs, to increase both labile, nutrient- 
supplying and stable pools of soil organic matter, or whether a single tree 
species can achieve both. 

How much organic matter does a soil need? 

For optimizing agroforestry techniques with respect to the maintenance 
of soil organic matter, it is important to know how much organic matter a 
given soil actually needs to remain fertile. The answer to this question 
certainly depends on the availability of technical substitutes for the 
different functions of organic matter in soil fertility, such as mineral 
fertilizer, irrigation, tillage and pesticides (van Noordwijk et al, 1997). Pieri 
(1989) discusses first steps towards defining critical soil organic matter 
contents for the maintenance of soil biological functions in sandy savanna 
soils but concludes that reliable values cannot yet be given. Minimum levels 
of total carbon or certain organic matter fractions in the soil to maintain 
an adequate soil structure are discussed in Section 10.5. 

86 C. Schroth ef al. 

4.2 Methods for Total Soil Organic Carbon 

Total organic carbon in soil is measured with wet or dry oxidation methods 
(Anderson and Ingram, 1993; Tiessen and Moir, 1993). Dry oxidation with 
automated CN-analysers has the advantage that all carbon forms in the 
soil are included in the measurement and that nitrogen and eventually 
other elements can be measured simultaneously. Procedures for wet 
oxidation in acid dichromate solution are available either with or without 
external heating. In the latter case, the oxidation of organic carbon is 
incomplete, and a correction factor is applied, such as assuming that only 
74% of the organic carbon has been oxidized. As the carbon recovery varies 
between soils in a usually unknown manner, this method should only be 
used for treatment comparisons with the same soil, but not across different 
soil types. Wet oxidation procedures largely exclude elemental carbon, 
such as charcoal, which can be an advantage in some situations, such as 
when studying the effect of management on organic matter in soils that 
contain large and/or variable amounts of charcoal. The importance of 
charcoal as a component of stable organic matter in soils is, however, 
increasingly recognized (see Section 4.1). Depending on the measurement 
method in wet oxidation and the combustion temperature in dry 
oxidation, carbonates are either included in the measurement (requiring 
acid pretreatment of the samples) or are not included (Tiessen and Moir, 

If total organic matter stocks of different soils or land-use types are to 
be compared, it is important to measure not only the carbon concentration 
in the soil at different depths, but also the bulk density for each depth 
interval, so that total quantities of carbon per unit area can be calculated. 
Land-use changes sometimes lead to a different distribution of organic 
carbon in the soil and litter profile - for example, as a consequence of soil 
tillage or of differences between trees and herbaceous vegetation in root 
distribution and litter decomposition. To distinguish such distribution 
effects from changes in total soil organic matter stocks, the soil should be 
sampled to at least 1 m depth or to the underlying parent material or a 
hardened horizon (Hamburg, 2000). If the areas to be compared differ in 
bulk density, as occurs where land use has caused pronounced soil 
compaction, sampling to the same depth implies that subsoil with low 
carbon content is included in the profile with the more compact soil, and 
correction may be necessary (de Moraes et al, 1996). 

4.3 Physical Fractionation Methods 

Physical fractions of soil organic matter can be obtained by separation into 
size classes (of primary particles or aggregates, see below), density classes 

Soil Organic Matter 87 

or with a combination of both techniques. Fractions of particular interest 
are the particulate organic matter (or macroorganic matter) and the light 
fraction because of their high activity and turnover rates (but see comments 
on these fractions in Section 4.1). Particulate organic matter can be defined 
as the fraction of 53-2000 (lm (sand fraction) that is obtained by sieving, 
as in normal particle size analysis, but without pretreatments for the 
destruction of organic matter, carbonates and iron oxides (Gregorich and 
Ellert, 1993). The lower limit of the fraction differs between authors, e.g. 
150 (lm (Meijboom et al., 1995) or 20 (lm (Feller et al., 1991a). The light 
fraction is normally obtained by density separation of soil in a liquid with 
a density of 1.4-2.0 g cm- 3 (several organic and inorganic liquids are in 
use), in which organic particles that are not associated with soil minerals 
float, whereas those associated with mineral components sink (Gregorich 
and Ellert, 1993; Stevenson and Cole, 1999). Some workers have combined 
these two approaches of separating an active soil organic matter fraction 
and have used the coarse sand fraction instead of the whole soil for density 
separation (Meijboom et al., 1995; Barrios et al., 1997). Anderson and 
Ingram (1993) recommended separation of the light fraction from the bulk 
soil <2 mm by flotation in water (for simplicity, instead of denser fluids) 
and collection of the floating material in a 0.25 mm sieve. When this 
method is used, care needs to be taken to standardize the water flow rate; 
moreover, dispersion even of sandy soils is not complete and fractions 
obtained do not consist solely of materials larger than 0.25 mm (B. 
Vanlauwe, unpublished results). With all fractionation methods, but 
especially when combining different fractionation techniques, care should 
be taken not to produce a large number of fractions which then cannot be 
related to pools with distinct properties in the soil. 

Authors have either dispersed the soil or not before physically 
separating soil organic matter fractions. Whether soil is dispersed and to 
what extent this is done depend on the purpose of the study. The major 
distinction is between whether the focus is on relationships between soil 
organic matter and primary mineral particles (sand, silt, clay), or on 
relationships between organic matter and soil structural properties, such 
as the physical protection of organic matter by macro- and micro- 
aggregates or effects of organic matter on the stabilization of the aggregate 
structure (Feller and Beare, 1997) (see also Section 10.5). For example, 
Six et al. (2000) used a wet sieving technique to separate aggregates and 
then dispersed the aggregates to quantify the free and the intra-aggregate 
particulate organic matter. Some soils are very difficult to disperse. Full 
dispersion of a Ferralsol to the level of primary particles, for example, may 
be both unpractical and undesirable. 

Soil dispersion can be accomplished by shaking with glass beads, use 
of ultrasound or by using chemicals such as sodium hexametaphosphate 
or sodic resins (Feller et al, 1991c). When using ultrasound, it is important 

C. Schroth ef al. 

not to apply too much energy as this may cause redistribution of carbon 
towards the lower particle size classes. Feller and Beare (1997) recommend 
using ultrasound for only the 0-50 [im soil fraction and not for the bulk 
soil (0-2 mm) to avoid such redistribution effects. The fraction >50 (1m 
can then be dispersed by shaking with glass beads or with chemical 
dispersants. Alternatively, a low amount of energy can be used for 
dispersing the bulk soil and then higher amounts for dispersion of the 
fractions (Amelung et al. , 1998). 

Chemical dispersion techniques may invalidate further 
characterization of fractions. For example, dispersing the soil with 
hexametaphosphate precludes later determination of phosphorus, so sodic 
resins can be used as dispersants in such studies (Feller and Beare, 1997). 
When using chemicals, it is important to standardize the shaking 
procedure on the basis of preliminary tests. For example, for a number of 
sandy savanna soils, the most complete dispersion with minimal 
redistribution of carbon towards the smaller particle size classes was 
achieved by shaking the soil in a sodium hexametaphosphate-sodium 
carbonate solution on a reciprocal shaker for 16 h at 144 rpm with a 
shaking amplitude of 4.5 cm (Vanlauwe et al., 1998b, 1999). To calibrate 
a certain dispersion procedure, organic matter with a size similar to 
particulate organic matter, or particulate organic matter extracted from 
another soil sample, can be added to the sample and its carbon content 
traced after dispersion. 13 C- or 14 C-labelled materials allow an accurate 
tracing of the added carbon. For an extensive review of physical 
fractionation approaches and results from tropical soils see Feller and 
Beare (1997). 

4.4 Chemical Methods 

The plant, faunal and microbial residues from which soil organic matter 
is derived consist of identifiable chemical substances such as carbohydrates, 
lignin, fats, waxes and proteins. These are decomposed and transformed 
in the soil through microbial and faunal action, and new substances are 
synthesized. The chemical composition and transformations of soil organic 
matter are studied with a number of specialized techniques, including 
chromatographic methods, nuclear magnetic resonance spectroscopy and 
analytical pyro lysis. These methods have been reviewed by Stevenson and 
Cole (1999) and Skjemstad et al. (1997), and recent results from tropical 
soils are given by Golchin et al. (1995), Skjemstad et al. (1997) and Zech et al. 
(1997). Results of such studies are potentially very relevant to agroforestry 
research, but their application is restricted to specialized laboratories. 

Easier to apply are techniques that attempt to separate labile fractions 
of soil organic matter through selective oxidation without chemically 

Soil Organic Matter 

characterizing these fractions. As mentioned in Section 4. 1, the separation 
of a labile organic matter pool through oxidation with a 333 mM solution 
of potassium permanganate (RMnOJ has been proposed by Blair et al. 
(1995). Modifications of the method, using less concentrated KMn0 4 
solutions, have been suggested by other workers. Bell et al. (1998) found 
that the carbon fraction oxidized with a 33 mM solution of KMn0 4 was 
more closely related to aggregate stability and (in combination with pH) 
effective cation exchange capacity of a range of Australian soils than that 
obtained with more concentrated solutions (for use of this fraction to 
predict rainfall infiltration and runoff see Section 10.5). Greater sensitivity 
of the fraction oxidized with a 33 mM than with a 330 mM KMn0 4 solution 
for detecting changes in the organic matter quality of cultivated soils in 
north-eastern Brazil was reported by Shang and Tiessen (1997). It should 
be noted that chemical fractionation of soil organic matter with 
permanganate destroys the respective fraction, whereas physical 
fractionation allows the further characterization of the isolated fractions. 
Soil organic matter has traditionally been separated according to its 
solubility in acid and alkali into humic acids, fulvic acids and insoluble 
humin (Anderson and Schoenau, 1993; Stevenson and Cole, 1999). 
However, these fractions are not closely related to soil organic matter 
functions and are not generally used in process-oriented research. 

4.5 Biological Methods 

For the measurement of soil microbial biomass, the fumigation-extraction 
method, which allows the simultaneous measurement of microbial nitrogen, 
phosphorus and sulphur, and various other methods are discussed by 
Voroney et al. (1993), Stevenson and Cole (1999) and Carter et al. (1999). 
Soil microbial biomass methods are prone to error when applying them after 
recent applications of fresh organic matter, which is a problem in 
agroforestry research. The magnitude of overestimation of microbial 
biomass after application of organic matter depends on the chemical 
characteristic that is measured to obtain microbial biomass. Vanlauwe et al. 
(1994) found that, when estimating microbial biomass from the 
concentration of soluble carbon in the soil before and after the fumigation, 
microbial biomass was substantially overestimated for at least nine days after 
residue application. When microbial biomass was estimated from ninhydrin- 
reactive nitrogen, overestimates were smaller and did not last as long. For 
nutrients in microbial biomass see the respective sections in Chapter 5. 

Soil respiration measurements provide an index of soil organic matter 
quality if the amount of carbon dioxide released is related to the total 
carbon present in the sample (Gregorich et al., 1994). Soil respiration can 
be measured in the field or under standardized conditions in the 

90 C. Schroth ef al. 

laboratory. The latter approach was used for comparing the organic matter 
quality in soils from different leguminous tree fallows (Schroth et al., 
1995b). In this study, the most sensitive indicator of species differences 
was the mineralization flush on the first day after rewetting the samples, 
which tended to increase with the litterfall of the tree species. Methods of 
separating root and microbial components of soil respiration are discussed 
by Hanson et al. (2000). 

The use of soil enzymes for assessing soil organic matter quality is 
discussed by Gregorich et al. (1994) and Stevenson and Cole (1999). For 
more detailed discussions of soil microbiological methods see Chapters 13 
and 15. 

Box 4.2. Identifying soil organic matter sources with carbon 

The amelioration of soil organic matter by trees is an important goal in agroforestry. 
Carbon isotope techniques can in some cases identify the source of soil organic 
matter by tracing the origin of its carbon. Plants that are cropped together or in 
sequence often possess different metabolic pathways for the assimilation of carbon 
dioxide from the atmosphere: most trees, legumes and root crops use the so-called 
C 3 pathway, whereas savanna grasses, maize and sorghum use the C 4 pathway. 
The metabolic pathway used by a plant affects the ratio of the two carbon isotopes, 
12 C and 13 C, in its biomass, because C 3 plants discriminate more against the heavy 
13 C isotope than C 4 plants and their biomass has therefore a lower 13 C: 12 C ratio. 
The carbon isotope ratios of the soil organic matter that is derived from this biomass 
vary accordingly. 

The stable isotopes 13 C and 12 C can be measured by dry combustion of a plant 
or soil sample and subsequent mass spectrometry. The 13 C: 12 C ratio of a sample is 
conventionally given in relation to a standard, usually the Pee Dee Belemnite 
(PDB), and expressed in parts per thousand (%o) as the 5 13 C value: 

5 13 C: 


12 r 

^ (sample) 


^ (sample) 

XlO 3 [%c 


Plants with the C 3 pathway have 8 13 C values in the range of -35 to -20%o, 
whereas plants that use the C 4 pathway have 5 13 C values between -1 9 and -9%o 
(Boutton, 1991). Because of the isotopic difference between C 3 and C 4 plants, 
carbon isotopes can be used for measuring the effect of vegetation change on soil 
organic matter, if either C 3 plants were introduced into an area that was previously 
dominated by C 4 species, or C 4 species were introduced in a previously C 3 

Soil Organic Matter 

dominated vegetation. If a C 3 species (e.g. a legume tree) is cultivated in a native 
C 4 vegetation (e.g. grassland), the contribution of the introduced tree to soil organic 
matter can be calculated as: 

f = (8 13 C - 5 13 C )/(8 13 C -5 13 C ) (4 2) 

'(C3) ^ U *-(C3) U ^-(C4)-"^ U *-(C3*) " *-(C4*y V*-*-) 

where f, C3 , is the proportion of soil organic matter derived from the C 3 tree species; 
5 ,3 C (C4) is the 5 13 C value of the soil under the original C 4 vegetation (control 
grassland plot); 5 13 C (C3) is the 8 13 C value of the soil under the introduced C 3 
vegetation; 8 13 C (C4 ,> and 8 13 C (C3 ,> are the 8 13 C values of the soil organic matter 
derived from the two plant species, respectively. The 8 13 C value for the soil organic 
matter derived from the original vegetation (8 13 C (C4 ,.) can be taken from the control 
plots covered by grassland. This is rarely possible for the introduced plant (S 13 C (C3 „,), 
because no fields exist at the same site which have been cropped with the tree 
species for the same time as the original grass vegetation. This value is usually 
substituted by the 8 13 C value of the plant material, either the aerial parts (for topsoil) 
or the roots (for subsoil) (Schweizer ef a/., 1 999). This is a valid approach, as the 
discrimination between the two carbon isotopes during mineralization and 
humification of plant material is believed to be negligible in the topsoil, although 
it may be of significance in the subsoil (Boutton ef a/., 1 998). 

The 8 ,3 C value can be affected by environmental factors such as water stress, 
but these effects are much smaller than the differences between C 3 and C 4 
vegetation (Boutton, 1 996). Therefore, the isotope composition of the soil organic 
matter in the topsoil reflects that of the living biomass or litter from which it is 
derived. Care has to be taken to remove carbonate, if present, from soil prior to 
analysis without changing the 8 ,3 C values of the soil organic matter (Midwood and 
Boutton, 1998). 

Carbon isotope methods have been used to assess the effect of land-use 
changes on soil organic matter quality (Feigl ef a/., 1 995), determine soil organic 
matter turnover rates (Bernoux ef a/., 1998), reconstruct vegetation history at the 
landscape scale (Boutton ef a/., 1 998), measure the contribution of root vs. shoot 
tissue to soil organic matter formation (Balesdent and Balabane, 1 996) and separate 
soil respiration into contributions from soil organic matter and added green manure 
(Nyberg ef a/., 2000). In an area in western Kenya that had previously been 
occupied by C 4 crops or pasture grasses, a comparison of 8 13 C values of topsoil 
under the canopy of different tree species (C 3 ) and outside the tree canopy detected 
significant effects of the trees on soil carbon as early as 5 years after tree planting 
(Nyberg and Hogberg, 1 995). Similarly, the contribution of scattered trees to soil 
organic matter in the Sahel savanna of Burkina Faso was quantified by comparing 
carbon isotope ratios in soil collected under the trees and in the open (Jonsson ef 
a/., 1999). Other agroforestry situations where the technique might be useful 
include rotations between C 4 crops and C 3 trees in fallow systems, and associations 
of C 4 crops and C 3 trees. A method for separating the effect of C 4 grasses and C 3 
legumes on soil organic matter in a mixed grass-legume pasture after rainforest 
has been developed by Cadisch and Ciller (1 996). 

Chapter 5 

Soil Nutrient Availability and Acidity 


1 Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil; 2 College of Agriculture and Life Sciences, Department of 
Crop and Soil Sciences, Cornell University, 909 Bradfield Hall, 
Ithaca, NY 14853, USA; 3 Centro Internacional de Agricultura 
Tropical (CIAT), AA 6713 Call, Colombia 

5.1 Synopsis 

In most tropical soils, the growth and production of crops is limited by the 
availability of one or several nutrients, such as nitrogen in sandy savanna 
soils and phosphorus in most acid soils (Ewel, 1986; von Uexkiill, 1986). 
Especially in humid climates, nutrient deficiency is often associated with 
soil acidity and aluminium toxicity, which may reduce the root 
development of plants and further aggravate the problems of insufficient 
nutrient supply from the soil. In dry climates, nutrient deficiency can limit 
the ability of crop and fodder plants to make use of the relatively short 
periods when there is sufficient soil moisture for plant growth (Breman 
and Kessler, 1997). 

Under favourable economic conditions, including the availability of 
credit, access to agrochemicals at reasonable prices, and a lucrative market 
for agricultural products, soil nutrient deficiency and acidity can be 
corrected through fertilizing and liming, and productive agriculture can 
then be possible even on previously very infertile soils. An example of this 
is the large-scale transformation of the acidic, nutrient-poor Oxisols of the 
Brazilian cerrado savannas into productive cropping systems through large 
inputs, especially of phosphorus fertilizer and lime (Sanchez, 1997). 

Many smallholder farmers in the tropics, however, can only afford to 
use small quantities of mineral fertilizers or sometimes none at all, and 
sufficient amounts of organic fertilizers, such as farmyard manure, are 
rarely available. In any case, with the exception of nitrogen fixed by 
leguminous fodder plants, the nutrients contained in manure would 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 93 

F.L. Sinclair) 

94 C. Schroth ef al. 

mainly constitute a redistribution of nutrients from other parts of the farm 
or landscape and not a true nutrient addition. Insufficient fertilizer 
applications to compensate for the nutrient exports from the farm in 
harvested products as well as nutrient losses from soil erosion, leaching 
and fire result in the widespread occurrence of negative nutrient balances 
in tropical farming systems (see Chapter 1). 

Nutrient efficiency 

As a consequence of low initial soil fertility and insufficient access to mineral 
and organic fertilizers to correct nutrient deficiencies and compensate for 
nutrient exports and losses, smallholder agriculture is widely practised 
under nutrient-limited conditions in the tropics. To be a viable land-use 
form under such conditions, agroforestry systems have to make efficient 
use of the nutrients available in the soil and of nutrient additions in mineral 
and organic fertilizers so that they are productive even if the nutrient 
supply is much lower than requirements for optimum growth. This should 
not be misunderstood to suggest that agroforestry systems could remain 
productive in the long term (i.e. be sustainable) without external nutrient 
additions to compensate for nutrient exports and losses, except - for a 
certain time - on very fertile soils or under conditions of very low yields. 
This is especially so for nutrients such as phosphorus that are only added 
to terrestrial ecosystems in very small quantities through atmospheric 
deposition or weathering of primary minerals (see Section 5.3). 

As a comparative measure of the ability of plant species or varieties to 
grow and produce yields on nutrient-limited soils, agronomists have 
developed the concept of nutrient efficiency (Marschner, 1995). A crop 
species or variety A is said to be more nutrient efficient than a species or 
variety B if A reaches a satisfactory level of productivity at a lower nutrient 
supply than B, even if both have the same productivity when nutrients are 
not limiting. Being more nutrient efficient, A can be grown on a less-fertile 
soil and with less fertilizer input than B. Differences in nutrient efficiency 
between plant species can be large: Marschner (1995) mentions three 
pasture species which needed 302, 87 and 26 mg P kg 1 of soil to produce 
90% of a common maximum biomass yield (this does not take into 
consideration eventual differences in fodder quality). The concept can also 
be applied to specific yield components, such as grain, instead of total 

Nutrient efficiency has two components: the efficiency of nutrient 
acquisition by the roots as measured by total nutrient uptake per plant, and 
the efficiency of nutrient utilization within the plant, which is called 
nutrient-use efficiency and is defined as the dry matter produced per unit 
nutrient in the dry matter. A large root system, efficient association with 

Soil Nutrient Availability and Acidity 95 

mycorrhizas and abundant nodulation in nitrogen-fixing plants are 
examples of characteristics that can lead to efficient nutrient acquisition. 
Efficient nutrient translocation within the plant and low nutrient 
requirements on the cellular level are factors that increase the nutrient- 
use efficiency. Alone or in combination, such characteristics tend to 
increase the productivity of plants when grown in soil with low nutrient 
availability (Marschner, 1995). 

Farmers with infertile soils will be interested in growing crop species 
with a high nutrient-use efficiency. For example, C 4 grasses make more 
efficient use of nitrogen and, therefore, have lower nitrogen concentrations 
in their tissue than C 3 grasses, and grasses in general make more efficient 
use of calcium and boron than dicotyledons. There are also differences 
between varieties (Marschner, 1995). In contrast to herbaceous crops, high 
nutrient-use efficiency is not generally a suitable selection criterion for 
agroforestry trees because, in addition to their production functions, the 
trees are often expected to increase nutrient cycling within the system, and 
a smaller amount of nutrients will be cycled if the trees have low nutrient 
concentrations. This function of agroforestry trees explains in part why 
the use of legume trees with their typically high nutrient concentrations 
is so common in tropical agroforestry. Suitable agroforestry tree ideotypes 
for situations where enhancing soil fertility is an objective would be both 
reasonably fast-growing and possess high tissue nutrient concentrations, 
which implies efficient mechanisms of nutrient acquisition rather than a 
particularly high internal nutrient-use efficiency. 

Effects of agroforestry practices on nutrient cycles 

The effects of agroforestry practices on nutrient cycles are the topic of 
several chapters of this book, and only a short overview is given here. A 
terminology for the analysis of nutrient cycles in agroforestry and other 
land-use systems is proposed in Box 5.1. 

Agroforestry practices can increase the total quantity of nutrients in 
the soil-plant system by increasing the nutrient transfers into the system 
and by reducing nutrient losses from the system. Increased nutrient inputs 
may originate either from the atmosphere or from soil compartments which 
are outside the reach of crop plants. The most important example for 
increased nutrient inputs from the atmosphere into agroforestry systems 
is biological nitrogen fixation by legume or other nitrogen-fixing trees (see 
Chapter 13). Trees can also increase atmospheric nutrient deposition but 
the quantities involved are usually small (see Chapter 9). If deep-rooting 
trees are associated or grown in rotation with shallow-rooting crops at a 
site where nutrients are available in deeper soil horizons, nutrients which 
were previously below the crop rooting zone may be made available to the 

96 C. Schroth ef al. 

crops via tree litter or primings. The same applies if tree roots reach 
laterally more distant soil and take up nutrients which are not accessible 
to the crop roots, e.g. in a stream bank or in a neighbouring plot or farm. 
Processes of nutrient capture that involve an increase in the effective 
rooting volume of a system through tree integration are discussed in 
Chapter 8. 

Agroforestry techniques can also help farmers avoid unproductive 
nutrient losses from a land-use system, i.e. losses of nutrients that are not 
exported in harvested products. Contour hedgerows and a continuous soil 
cover reduce soil erosion and surface runoff on sloping land (see Chapter 
17), and the permanent or intermittent presence of deep tree roots can 
help to reduce nutrient leaching, especially during times of the year when 
no crops are present (see Chapter 7). Permanent associations of crops with 
trees also oblige the land user to protect the field from fire and associated 
nutrient losses (see Chapter 9), although tree planting in fallow rotations 
does not necessarily have this effect (MacDicken, 1991). 

Trees could also increase the total amount of nutrients cycling in a 
system by accessing nutrient pools that are not accessible to crops, for 
example through more efficient mycorrhizal associations (see Chapter 15) 
or solubilization of recalcitrant phosphorus forms in the rhizosphere (see 
Chapter 15). However, such effects are not yet well researched. 

Box 5.1. A terminology for studying nutrient cycles of 
agroforestry systems. 

For the conceptual analysis, experimental study and modelling of nutrient cycles 
in agroforestry systems, it is useful to subdivide the total nutrient quantity within 
a system into pools and compartments as illustrated in Fig. 5.1 . 

Nutrient pools 

A nutrient pool is defined here as a fraction of the total amount of a given nutrient 
in the plant-soil system whose chemical, physical or biological properties 
differentiate it from other fractions of the same nutrient with respect to reactivity, 
mobility and availability to plants, animals or microorganisms. In an agricultural 
or agroforestry system, the nutrients in the biomass, necromass and soil constitute 
separate pools, and each of these can be further subdivided as needed for the 
purpose of a study. In particular, the different forms of a nutrient in the soil are 
often subdivided into various pools, commonly organic vs. inorganic and labile 
vs. stable (Fig. 5.1). The differentiation between pools of the same nutrient is 
sometimes clear-cut and sometimes rather arbitrary and dependent, for example, 
on the extraction method used. 

Soil Nutrient Availability and Acidity 





p Litter layer 











Tree litter, 
crop residues, 



Dead roots, 
dead fauna, 
dead microbes 






Fixed P, 
K, NH„, 




Deep roots, 
some fauna 

Dead roots, 
dead fauna 








A rt\&^ Biomass 

Necromass Water 

Soil Soil Soil Soil 

org. /stable org. /labile inorg. /stable inorg. /labile 


Fig. 5.1. Schema of the subdivision of total nutrient stocks in agroforestry 
systems into pools and vertical and horizonal compartments according to their 
physicochemical form and location, org., organic; inorg., inorganic. 


In contrast to pools, which differentiate between physicochemical or biological 
forms of a nutrient, nutrient compartments subdivide the total quantity of nutrients 
within a system according to their location (Fig. 5.1 ). For example, the subdivision 
of the soil profile into topsoil and subsoil is useful in an agroforestry study if the 
topsoil is defined as the rooting zone of shallow-rooted crops, and the subsoil is 
defined as the soil not accessible to these crops but still within the reach of trees. 
Nutrients in the shoots and roots are assigned different compartments, because 
they differ in their susceptibility to loss from the system by harvest or fire and their 
ease of management through, for example, pruning. A further subdivision of the 
system into horizontal compartments is useful in simultaneous agroforestry 
systems. Specifically, nutrients in the soil directly under trees and crops may be 
distinguished from each other and from nutrients at a distance too far to be reached 
by crop roots but accessible to tree roots that have greater lateral extension. The 
most important criterion for the distinction of both pools and compartments is that 
they are useful for the conceptualization, measurement and modelling of the 
nutrient cycles in the plant-soil system as influenced by land-use practices. 


C. Schroth ef al. 

Box 5.1 . Continued. 

Nutrient transfers 

A nutrient transfer is a physical movement of a nutrient into or out of a system 
(typically a plot or a farm) or between different compartments within a system. 
Examples of nutrient transfers across the plot boundary include imports of nutrients 
with rainfall or dust and nutrient losses with fire. Leaching of a nutrient from the 
topsoil into the subsoil or nutrient returns from the vegetation to the soil via 
throughfall, stemflow and litterfall are examples of nutrient transfers between different 
compartments within a system. Nutrient transfers are often called fluxes but this term 
is also commonly used to refer to exchanges among pools (see Box 4.1 ). 

Nutrient transformations 

Nutrient transformations are exchanges of nutrients between pools. Unlike nutrient 
transfers, they do not involve a major physical movement of the nutrient, but rather 
one or several biochemical or physicochemical reactions that change the reactivity, 
mobility and/or availability of the nutrient. Examples of nutrient transformations 
include the mineralization or immobilization of nitrogen, phosphorus and sulphur 
in litter and soil, or exchanges between physicochemical forms of phosphorus and 
potassium which differ in their reactivity and availability to plants. 

Nutrient transformations and transfers often affect each other. For example, 
nitrification in the topsoil (a transformation) strongly increases nitrogen leaching 
into the subsoil compartment (a transfer). 

Nutrient efficiency and tree-crop interactions 

Through the mechanisms mentioned above, the total amount of nutrients 
that are taken up from soil and air by a tree-crop association may be larger 
than that available to a crop alone. If the integration of deep-rooting, 
nitrogen-fixing, highly mycorrhizal trees into a field with annual or 
perennial crops increases the total nutrient uptake and biomass production 
of a land-use system on a nutrient-poor soil, then the practice has been 
successful in terms of improved nutrient efficiency. However, it can still 
be an economic failure. Although the trees may increase the acquisition of 
certain soil resources and the biomass production of the system, they may 
also compete with the crops for the same and other resources, and the 
overall economic output of the association may be less than that of the crop 

The partitioning of soil resources between trees and crops, and 
therefore their respective productivity in an association, depends on the 
interactions between the associated species. These interactions can be 

Soil Nutrient Availability and Acidity 99 

complex, involving a large number of processes and their spatiotemporal 
patterns, and are rarely fully understood for a given association. For their 
conceptualization and analysis, it is useful to distinguish between three 
principal types of interactions: 

• with competition, uptake of a nutrient by species A reduces its 
availability for species B; 

• with complementarity, uptake of a nutrient by species A does not affect 
its availability for species B, so that nutrient uptake by species A and 
B is additive; 

• with facilitation, species A increases the availability of the nutrient for 
species B. 

Details and examples for these types of interactions are given in Box 

In simultaneous agroforestry systems, these types of interactions rarely 
or never occur in isolation in the field. For example, a deep-rooting tree 
may utilize leached nitrate in the subsoil that is not accessible to an 
associated annual crop, and the interaction between the two species with 
respect to this nitrate would then be complementary and beneficial for the 
system as a whole. However, the tree would also acquire some of its 
nitrogen as well as other nutrients and water from the top soil, and here 
the interaction with the crop could be competitive. Competition in the 
topsoil could even be an important incentive for the tree to form deep 
roots and therefore a precondition for complementarity between the two 
species in the subsoil. Similarly, the integration of nitrogen-fixing trees 
into a cropping system on a nitrogen-poor soil can increase the total 
nitrogen acquisition by the system, and the trees could have a facilitative 
effect on the crops through the release of nitrogen from tree primings, 
but could at the same time compete with the crops for other nutrients, 
water, light or simply space. 

Whether an association is successful or not depends on the balance 
between competitive, complementary and facilitative interactions on the 
one hand, and on the relative value of tree and crop products on the other 
hand (van Noordwijk and Purnomosidhi, 1995). The higher the value of 
the trees, the more competition with the crops will be accepted, and the 
lower their value, the more the resource use by trees and crops needs to 
be complementary or facilitative. In many agroforestry associations, the 
(short-term) benefit from the crops will be far higher than that from the 
trees, and little competition will therefore be tolerated. Besides requiring 
time and sometimes monetary inputs, in most cases tree planting reduces 
the space available for the crops. Therefore, planting trees with no direct 
economic value (e.g. certain legume trees) can only be economically 
advantageous if they exert a net facilitative effect on the crops. 
Furthermore, this effect would have to become perceptible over a relatively 

100 C. Schroth etal. 

short time period to provide an incentive for tree planting (see Chapter 
2). Failure to take these considerations into account may help to explain 
the low adoption of some agroforestry techniques, such as alley cropping. 

Box 5.2. Types of interactions between plants. 

There are many ways in which plants growing on the same piece of land can 
interact. Of particular relevance for understanding and managing nutrient and 
water cycles in agroforestry are three types of interactions: competition, 
complementarity and facilitation. In agroforestry systems, these types of 
interactions generally occur simultaneously and may influence each other (see 


A major problem in agroforestry associations is competition between trees and 
crops for light, water and nutrients. Competition for a nutrient occurs when two 
species attempt to meet their demand for this nutrient from the same pool and 
compartment in the soil (see Box 5.1 ), and when their combined demand exceeds 
the supply during a certain time interval. Some competition is probably inevitable 
where overall resource use is efficient (for example, to reduce nutrient leaching, 
an intensive, interlocked root system is needed), but excessive competition leads 
to the suppression of plants and low yields. It is, therefore, a general objective in 
agroforestry to avoid strongly competitive relationships between species and to 
create complementary or even facilitative relationships. 

There are several conceptually important distinctions within the term 
competition. Intraspecific competition occurs between individuals of the same 
species, and interspecific competition between individuals of different species. 
Everything else being equal, the former is commonly assumed to be more intense 
than the latter because individuals of the same species have exactly the same 
resource requirements and uptake patterns (i.e. they occupy the same niche). 
Exploitation competition occurs when plants compete directly for the same 
nutrients or water in the soil, i.e. when the nutrient or water depletion zones around 
their roots overlap. Interference competition, on the other hand, occurs when one 
individual indirectly impedes the access of the other individual to the limiting 
resource, for example by producing allelopathic substances which reduce root 
growth of competing species. Finally, size-symmetric competition describes the 
situation when two competitors obtain the contested resource in proportion to 
their size, whereas, with size-asymmetric competition, the larger plant obtains a 
disproportionate share of the resource. Competition for light is usually strongly 
asymmetric (a plant needs to be only a little taller than its neighbour to obtain the 
lion's share of the incoming radiation), whereas competition for nutrients seems 
to be more symmetric. Little is known about the symmetry of competition for water 
(Schwinning and Weiner, 1998). 

Soil Nutrient Availability and Acidity 101 


Complementarity between two interacting species means that the use of a resource 
by one species is not at the expense of the use of this resource by the other species. 
With respect to soil nutrients this occurs when two species use different pools or 
compartments of a nutrient, or when they use the same pool or compartment at 
different times. This is a desirable situation, because the two species in 
combination would then tend to use this nutrient more effectively than either 
species alone. Several examples for complementary nutrient use in agroforestry 
systems have been mentioned in the text: capture of nutrients from the subsoil or 
laterally distant soil by trees, differential access to atmospheric nitrogen in 
associations of nitrogen-fixing and non-nitrogen-fixing species, and differences in 
the access to recalcitrant nutrient pools in the soil. Complementarity through the 
temporal separation of nutrient uptake would be expected when associated species 
differ in their phenology and more markedly in rotational systems when plants are 
deliberately grown at different times. 


Facilitation means that one species actually improves the growth conditions of the 
other species. There is a substantial literature on facilitative interactions in natural 
plant communities (Callaway, 1995; Callaway and Walker, 1997). In fallow 
rotations, facilitation of crop growth by the fallow trees is the expected 
consequence of the accumulation of nutrients, including fixed nitrogen, in the 
biomass during the fallow phase and their release following the reconversion of 
the fallow into a crop field. Facilitation can also be achieved through the 
application of nutrient-rich tree prunings to crops in a tree-crop association, 
provided that the nutrients released from the prunings are predominantly taken up 
by the crops and not reabsorbed by the trees themselves. 

Optimization of nutrient cycles in agroforestry 

The tools that are available to a farmer for optimizing nutrient cycles are 
the selection of plant (and animal) species, their spatial and temporal 
arrangement (system design), and their management. 

The selection of crop species, be they herbaceous or perennial, is 
generally dominated by market and household constraints, but 
considerable flexibility often exists with respect to the selection of service 
trees that are grown in association with them. Tree species differ with 
respect to their growth rates (and thus nutrient uptake), root distribution 
(and thus access to subsoil nutrients), litter production (and thus nutrient 
return to the soil), litter quality (and thus decomposition dynamics), ability 

102 C. Schroth etal. 

to fix nitrogen and many other characteristics that affect nutrient cycles. 
As mentioned before, high nutrient-use efficiency may be an economically 
important selection criterion for crop species and varieties in agroforestry 
systems, but may not be a good guide for selecting tree species, because 
the production of nutrient-rich, easily decomposable litter is a desirable 
property even for trees that serve predominantly production purposes. 
However, trees with high nutrient-use efficiency (i.e. low tissue nutrient 
contents) and high lignin and polyphenol contents could be useful for soil 
organic matter build-up (see Chapter 4). 

Nutrient cycles are greatly affected by the way trees and crops are 
arranged in space and time (system design). Trees can be planted or 
allowed to regenerate in association with crops, as in shaded coffee and 
cocoa plantations, parkland systems, contour hedgerows and boundary 
plantings, and in each case the relative access of trees and crops to 
nutrients in the soil can to some extent be influenced through tree spacing 
and planting design. Trees can also be grown in rotation with crops, as in 
fallow systems. Trees may enhance nutrient cycling in both simultaneous 
and rotational systems. However, in the former case, they access at least 
partly the same nutrient pools and compartments in the soil as the crops 
(see Box 5.1) and may cause competition, although their net effect may 
still be beneficial because of microclimatic protection of shade-demanding 
crops, timber production, soil conservation or biological nitrogen fixation. 
In fallow systems, in contrast, trees and crops take up nutrients from the 
same soil at different times. Improved fallows are, therefore, seen as a 
means of minimizing competition while maximizing facilitative effects of 
trees on subsequent crops. Some competition between fallows and 
neighbouring crop plots may, however, occur through lateral tree roots 
in the small-scale patchwork of fallows and crop fields that is typical for 
tropical smallholder agriculture (van Noordwijk, 1999). Fallow systems 
permit the use of fast-growing tree species that would be too competitive 
for use in simultaneous tree-crop associations (Schroth et al., 1996), and 
these can be planted at narrow spacing to achieve rapid biomass 
accumulation, soil regeneration and weed suppression without the need 
to be held in check by periodic pruning. 

Agroforestry systems can be managed so as to reduce competitive 
interactions and increase complementary and facilitative interactions 
between associated tree and crop species (Lehmann, 2002). However, care 
is needed in deciding what measures to apply and how to apply them. 
Periodic shoot pruning of trees and transfer of the biomass to crops, as in 
tree crop plantations shaded by legume trees or in hedgerow 
intercropping, reduces nutrient sequestration in the tree biomass, 
improves the quality of the tree biomass (more and younger leaves, less 
old wood) and makes nutrients available to the crops, beside reducing the 
competition of the trees for water and light. Frequent pruning of trees 

Soil Nutrient Availability and Acidity 103 

during the rainy season can cause an 'anti-cyclic' root development of the 
trees by shifting the period of maximum tree root growth into the dry 
season, thereby increasing the potential for complementarity with annual 
crops in the use of soil resources (Schroth and Zech, 1995b). However, the 
root systems of frequently pruned trees have also been found to become 
shallower (see Section 8.1), and nitrogen fixation of legume trees may be 
impeded by frequent, intensive shoot pruning (see Section 13.1). Similarly, 
soil tillage at the beginning of the cropping season destroys tree roots in 
the topsoil and gives the developing crop a temporary advantage in the 
competition for soil resources, but may also provoke increased 
mineralization of soil organic matter and nutrient release at a time when 
the crops are too small to absorb these nutrients. 

Matching agroforestry practices to sites 

As mentioned in Chapter 1 , it is important to keep in mind that tropical 
sites and soils differ widely in their fertility, including the availability of 
different nutrients, and that agroforestry practices differ in their potential 
to adapt to, and eventually to mitigate, certain fertility problems. This 
means that not only crop and tree species, but also agroforestry techniques 
have to be matched to sites according to their respective potentials and 
limitations. For example, on sandy savanna soils, low nitrogen availability 
is often a limiting factor for crop production, indicating that agroforestry 
with nitrogen-fixing trees can contribute to improved soil fertility and 
higher yields (Barrios et al., 1997). For certain forest soils, in contrast, 
nitrogen is less limiting, and the introduction of nitrogen-fixing species 
would not be given high priority (Schroth et al, 1999a). On shallow soils, 
which are common in West Africa, root competition between trees and 
associated crops can be particularly intensive. Under these circumstances, 
simultaneous agroforestry systems may only be viable if uncompetitive tree 
species are used (Schroth and Lehmann, 1995) and improved fallows may 
be preferable. At sites where nutrient accumulation in the subsoil has been 
detected, agroforestry techniques can be specifically designed for the 
recycling of these nutrients, for example, through high-density plantings 
of fast-growing tree species as improved fallows (Jama et al., 1998). 

It should also be mentioned that in the tropics, agroforestry techniques 
are used not only under nutrient-limited conditions, but also in systems 
with relatively high fertilizer inputs, such as certain shaded coffee 
plantations (Beer et al, 1998). Under such conditions, trees can still have 
beneficial effects on nutrient cycles, especially through soil protection and 
the reduction of nutrient leaching (Schroth et al, 2001b). 

104 C. Schroth etal. 

5.2 Methods for Soil Nitrogen (E. Barrios, G. Schroth) 

The prominent role that nitrogen occupies among the nutrient elements 
is because of the relatively large amounts of it that are required by plants 
and soil microorganisms in comparison with other nutrients. Low soil 
nitrogen availability limits crop growth on many tropical soils, especially 
those with low organic matter contents such as sandy soils and soils of 
semiarid environments. On such soils, the availability of nitrogen to crops 
can be effectively increased through the use of nitrogen-fixing trees in 
tree-crop associations or improved fallows. Nitrogen deficiency in plants 
is readily recognized by yellow coloration of the older leaves and by slow 
and stunted growth (Stevenson and Cole, 1999). 

More than 95% of the nitrogen in topsoil is usually present in organic 
forms. Through microbial mineralization it is transformed into ammonium 
(NH 4 +) and then further oxidized by nitrifying soil microbes to nitrite 
(N0 2 ~) and nitrate (N0 3 ~), a process called nitrification. Recent results 
suggest that the opposite process, the dissimilatory reduction of nitrate to 
ammonium by microbes, which is usually assumed to occur only under 
flooded conditions, may be a relevant process also in upland tropical forest 
soils (Silver etal, 2001). 

Plants absorb most of their nitrogen from the soil as ammonium or 
nitrate, with nitrite usually being present only in minor quantities in well- 
aerated soils. However, both trees and herbaceous plant species have the 
ability to take up certain organic nitrogen forms, especially when they are 
associated with mycorrhizal fungi (Turnbull et al., 1995; Nasholm et a!,., 

Burning of vegetation as in shifting cultivation can lead to large losses 
of biomass nitrogen (see Section 9. 1). As nitrate, nitrogen is also easily lost 
from agricultural soils through leaching, especially following the 
mineralization peak that typically occurs at the beginning of the rainy 
season in seasonally dry climates (see Section 7.1). Recycling of leached 
nitrate from the subsoil can be an important function of trees in 
agroforestry systems (see Section 8.1). Nitrogen losses also occur through 
denitrification (Granli and Bockman, 1994; Barton et al, 1999) and 
volatilization of ammonia, for example from decomposing biomass or 
surface-applied fertilizer, especially urea. 

The amount of mineral nitrogen that is available to a crop during its 
life cycle can be calculated as the mineral nitrogen present in the soil at 
the beginning of the cropping season, plus additions and minus losses 
during the cropping season. The additions include those from nitrogen 
fixation, where this occurs (see Chapter 13), mineralization of soil organic 
matter (see below), release from crop and tree residues (see Chapter 6), 
atmospheric deposition (see Chapter 9) and fertilizer application. Losses 
occur through nitrate leaching (see Chapter 7), volatilization and 

Soil Nutrient Availability and Acidity 105 

denitrification. The present section concentrates on the quantification of 
the different forms of nitrogen in the soil and nitrogen mineralization. 
More detailed information on the other processes involved in nitrogen 
cycling in agroecosystems are given in the previously indicated sections of 
this book and in reference texts such as Stevenson and Cole (1999) and 
Havlin et al. (1999). A recent review of nitrogen management in African 
farming systems has been provided by Giller et al. (1997). Methods for 
using 15 N to track nitrogen movements in the plant-soil system are 
discussed in Sections 7.3 and 8.2. 

Total nitrogen 

The methods for the measurement of total nitrogen in soil and plant 
materials are based either on wet oxidation or dry combustion. The dry 
combustion method is used in automated CN-analysers (see also Section 
4.2). Their advantage is the inclusion of all mineral and organic nitrogen 
forms and the simultaneous measurement of carbon and sometimes other 
elements. Because of the small sample sizes for certain instrument types, 
careful homogenization and fine grinding of the soil is necessary. For 15 N 
studies, CN-analysers can be directly connected to an isotope mass 
spectrometer (McGill and Figueiredo, 1993). The commonly used wet 
oxidation method is based on the Kjeldahl digestion procedure with 
sulphuric acid and a catalyst (Anderson and Ingram, 1993). A problem 
with this method is that some, but not all of the nitrate in soil is included 
in the measurement, which precludes the separate quantification of nitrate 
and addition of the value to the total nitrogen value from the Kjeldahl 
analysis. For soils containing relevant amounts of nitrate and nitrite, 
pretreatments are available that first convert nitrite into nitrate using 
permanganate and then nitrate into ammonium using reduced iron, which 
is included in the Kjeldahl analysis. These pretreatments are also 
recommended in studies with 15 N tracers because of the influence of highly 
labelled nitrate and nitrite on the tracer analyses (McGill and Figueiredo, 

Mineral nitrogen 

At the beginning of a cropping season, the soil may contain significant 
amounts of mineral nitrogen, including residual nitrogen from the 
previous crop and nitrogen that was mineralized between the two crops 
(e.g. during the dry season and especially upon rewetting of the soil at the 
beginning of the rainy season). Mineral nitrogen in the soil is usually 
measured by extracting field-moist soil samples with a 1 M or 2 M solution 

106 C. Schroth etal. 

of potassium chloride and colorimetric measurement of ammonium and 
nitrate (+ nitrite) in the extract, with the common assumption that nitrite 
is of little importance in well-aerated soils. A detailed procedure is given 
by Maynard and Kalra (1993). 

Preseason mineral nitrogen can make a substantial contribution to the 
total nitrogen supply of a crop. In experiments with legume tree fallows 
on a sandy, nitrogen-deficient soil in Zambia, it was the most sensitive 
measure of plant-available nitrogen and explained close to 50% of the 
variance in crop yields (Barrios et a!,., 1998). Soil nitrate alone has been 
widely used as a measure of nitrogen availability in temperate soils because 
it correlates with crop yield, is strongly influenced by soil management and 
has good reproducibility (Stanford, 1982). However, in tropical soils with 
a pronounced dry period, ammonium can be seasonally important because 
it can accumulate during the dry season (Wong and Nortcliff, 1995). Its 
inclusion in measurements of mineral nitrogen in tropical soils is generally 
recommended, unless it has been established that ammonium 
concentrations are small and more or less constant in a given soil. Mineral 
nitrogen accumulation in the soil as related to nitrogen leaching and 
recycling by trees is discussed in Box 8.1. 

Nitrogen mineralization 

Topsoils may contain several thousand kilograms of nitrogen per hectare, 
most of which is locked up in soil organic matter and not directly available 
to plants. The microbial transformation of organic into mineral nitrogen, 
or nitrogen mineralization, is a process of fundamental importance for the 
nitrogen supply in both natural and agricultural ecosystems. Net nitrogen 
mineralization is the difference between the total (or gross) nitrogen 
mineralization through microbial decomposition of organic matter and 
the simultaneous immobilization of nitrogen in microbial biomass. Gross 
rates of nitrogen mineralization can be determined with 15 N techniques 
(Barraclough, 1995). In the following discussion, nitrogen mineralization 
is taken to mean net mineralization. 

Differences in the quantity and quality of soil organic nitrogen, climatic 
factors such as temperature and moisture, and soil management such as 
tillage and incorporation of organic matter, lead to considerable spatial 
and temporal variability in nitrogen mineralization rates in the field. 
Nitrogen mineralization can be measured in the field or under 
standardized conditions in the laboratory, or it can be predicted using 
models of variable complexity from knowledge of the soil type, climate 
and management. The prediction of nitrogen mineralization in the field 
is an important research issue in temperate, high-input agriculture 
because of the danger of nitrate leaching into the groundwater if the 

Soil Nutrient Availability and Acidity 107 

supply of soluble nitrogen exceeds crop requirements (Campbell et al., 

Of particular interest for tropical agroforestry research is the 
measurement of nitrogen mineralization rates as related to soil 
management practices such as biomass application, fallow type and length, 
and method of soil tillage (Barrios et al., 1996). The analysis of temporal 
and spatial patterns of nitrogen mineralization can provide important 
information for optimizing the spatiotemporal correspondence between 
nitrogen supply in the soil and its uptake by plants, for example through 
early sowing in soils with a large mineralization flush at the beginning of 
the rainy season, fertilizer application during periods with insufficient 
mineralization, or increased planting density in locations with high 
mineralization rates. The concept of synchrony and synlocation between 
nutrient supply and demand is further discussed in Section 6.1. Small- 
scale patterns of nitrogen mineralization within agroforestry plots can also 
be used as a sensitive measure for the effect of different plant species and 
management factors on nitrogen availability (Schroth et al. , 2000a, 2001c). 

Nitrogen mineralization is measured in the field by extracting the 
mineral nitrogen (ammonium and nitrate) both at the beginning and at 
the end of an incubation period. Ideally, this is done with undisturbed soil, 
as soil disturbance may increase mineralization rates (Stenger et al. , 1995). 
Following a procedure devised and tested by Raison et al. (1987), Anderson 
and Ingram (1993) recommended the use of iron or plastic tubes for 
isolating a certain soil volume and preventing uptake of mineral nitrogen 
by roots. The tubes are driven into the soil and are covered to prevent 
leaching of mineral nitrogen by rain. For every site or treatment, three 
tubes are removed at the beginning and three other tubes at the end of 
the incubation (after about 1-2 weeks), and mineral nitrogen is extracted 
from the bulked replicates. Variants of the method involve placing resin 
bags at the top and bottom of the tubes in order to remove mineral 
nitrogen from rainwater and percolating soil solution (Zou et al. , 1992), or 
placing the undisturbed soil samples in sealed plastic bags, the buried bag 
method (Zou etal., 1992; Barrios and Herrera, 1994; Subler et al. , 1995). 

As cut-off roots cannot be removed before incubating undisturbed 
samples, errors in nitrogen mineralization measurements could occur 
because coarse woody roots continue taking up nitrogen, or because 
nitrogen is immobilized during microbial decomposition of root tissue 
(Raison et al, 1987). Bauhus (1998) did not find indications for the latter 
process when using the method in a European beech forest (microbial 
biomass nitrogen was not increased in the incubated samples), although 
consistent negative mineralization rates in pasture in Brazil and woody 
vegetation in Zimbabwe indicate that microbial immobilization of nitrogen 
is a problem where large amounts of poor-quality organic matter are 
present (K.E. Giller, unpublished data). 

108 C. Schroth etal. 

The variability of net nitrogen mineralization rates measured in 
undisturbed soil cores is often large (Raison et al., 1987). Using disturbed 
samples for the incubation can reduce the variability, because soil from 
several sampling positions can be mixed, and the sample for the 
determination of the initial mineral nitrogen content of the soil can be 
taken directly from the soil sample to be incubated. Coarser roots can be 
removed from the soil before the incubation. Plastic bags have been widely 
used for nitrogen mineralization experiments with disturbed soil. This is 
also called the buried bag method in the literature, so it is important to 
specify whether disturbed or undisturbed soil was used. If the method is 
used with disturbed soil, it is recommended that net nitrogen 
mineralization rates in disturbed soil are compared with those in 
undisturbed soil cores in preliminary experiments to ascertain the effect 
of disturbance on mineralization rates and to determine appropriate 
correction factors. In Amazonian forest soils, good agreement between 
nitrogen mineralization rates in disturbed and undisturbed soil samples 
has been found (Piccolo et al., 1994; Schroth et al., 2001c), although 
the agreement was not as good in soil from pasture (Piccolo et al, 1994). 
In certain situations, penetration of tree roots into the incubated plastic 
bags or perforation of the bags by soil fauna can cause problems. The 
former problem can be prevented by periodic trenching around the 
incubated bags, and the latter by placing the bags between sheets of 
mosquito netting. 

Measurements of nitrogen mineralization under standardized 
conditions in the laboratory are used for relative comparisons between 
soils and treatments and to determine input parameters for predictive 
models of nitrogen mineralization in the field (e.g. Campbell et al., 1995). 
Both aerobic and anaerobic incubation methods are available. Depending 
on the exact research objective, disturbed or undisturbed soil samples can 
be used (i.e. excluding or including interference from soil physical 
properties), and factors such as temperature and moisture can be varied 
as required. If the soils were dried prior to the incubation, a mineralization 
flush usually occurs upon rewetting, which should be quantified separately 
from the subsequent mineralization rate, as treatment effects may differ 
between these two phases of mineralization (Schroth et al., 1995b). 
Laboratory incubations have been used to compare the nitrogen 
availability of different fallow soils and to identify those characteristics of 
fallow tree species that are associated with high nitrogen availability in the 
soil. In a sandy savanna soil with low total nitrogen content (0.7 g kg 1 ) in 
Zambia, nitrogen mineralization in the soil was highest after nitrogen- 
fixing trees with a low (lignin + polyphenol) :N ratio in the leaves, 
indicative of rapid decomposition of the biomass (Barrios et al, 1997) (see 
also Section 6.1). In contrast, in a forest soil in Cote d'lvoire with higher 
total nitrogen contents (2.3 g kg 1 ), soil nitrogen mineralization and crop 

Soil Nutrient Availability and Acidity 



— 65 


- 60 

g> 55 


50 - 

2 45 - 

i 40 


30 4 



Albizia guachapele 
Albizia niopoides 
Leucaena collinsii 
# Leucaena leucocephala 
A Ateleia herbert-smithii 
■ Gliricidia sepium 
O Caesalpinia velutina 
<3> Caesalpinia coriaria 
A Caesalpinia eriostachys 
V Control 

0.0 0.5 1.0 1.5 2.0 

Root-C (Mg ha" 1 10 cm" 1 ) 



Fig. 5.2. Increase of nitrogen mineralization in the soil with increasing root mass of 
six nitrogen-fixing fallow tree species on a Lixisol in Cote d'lvoire. The three 
Caesalpinia species and the spontaneous fallow (control) dominated by 
Chromolaena odorata were not nitrogen fixing (reproduced with permission from 
Schroth etal., 1995b). 

yields increased with litterfall and root mass of the fallow trees, but were 
not influenced by the ability of the trees to fix nitrogen (Figure 5.2). For 
detailed procedures of laboratory incubation methods see Anderson and 
Ingram (1993), Campbell et d. (1993) and Hart et d. (1994). 

Fractionation of organic nitrogen: the potentially mineralizable pool 

Considering the importance of nitrogen mineralization for plant growth 
and yields on most soils, it is not surprising that considerable efforts have 
been made to identify and quantify the labile, potentially mineralizable 
pool of soil organic nitrogen. On the one hand, these efforts aim at the 
development of easily measurable indices for the prediction of nitrogen 
mineralization in the field, which could replace lengthy incubation studies. 
On the other hand, an improved understanding of the physicochemical 
characteristics of labile soil nitrogen would facilitate the study of 
management effects on nitrogen availability and specifically the 
development of measures to increase the pool of readily available nitrogen 
in nitrogen-limited soils. The search for labile and stable pools of organic 

110 C. Schroth etal. 

nitrogen is also related to computer models designed to predict soil organic 
matter and nitrogen dynamics, which divide total soil organic matter or 
nitrogen into two or more pools with different turnover rates, but 
generally suffer from the inability to quantify these functional pools 
directly (see Box 4.1). 

Physical fractionation methods 

Several studies have shown that the addition of nitrogen-rich biomass to 
soil can lead to an increase of nitrogen in labile soil organic matter, and 
consequently increased nitrogen mineralization in the soil and nitrogen 
supply to crops (Barrios et al, 1997). Depending on the soil and the 
quantity of biomass nitrogen applied, the effect may, however, become 
significant only after several years (Haggar et al., 1993). In research with 
different planted tree fallows on a severely nitrogen-limited, sandy savanna 
soil in Zambia, the physical fractionation of soil organic matter according 
to particle size and density was found to be a sensitive indicator of the 
effects of nitrogen-rich biomass on nitrogen availability in the soil (for 
physical fractionation of soil organic matter see Section 4.3). Two- and 3- 
year-old planted tree fallows with Sesbania sesban and Gliricidia sepium 
increased the amount of nitrogen in the light particulate organic matter 
fraction (> 150 |im, <1.13 g cm 4 ), which correlated with nitrogen 
mineralization of the whole soil. Soil inorganic nitrogen and subsequent 
maize yields were consequently increased in these fallow treatments 
compared with continuous unfertilized maize (Barrios et al, 1996, 1997). 

In a study with 13 N-labelled tree residues on a Nigerian Lixisol, 
significant relationships between the uptake of residue-derived nitrogen 
by maize and residue-derived nitrogen in the particulate organic matter 
(>53 |im) also confirmed the value of this fraction as an indicator of high 
nitrogen availability (Vanlauwe et al., 1998b). In a set of alley cropping 
trials in the West African moist savanna, the relative change in nitrogen 
in particulate organic matter was about twice the relative change in total 
soil nitrogen content, and the proportion of the total soil nitrogen in 
particulate organic matter increased significantly with annual nitrogen 
inputs in crop residues and tree prunings. This also suggests that nitrogen 
incorporated in this fraction is relatively labile and a more sensitive 
indicator of management effects than total soil nitrogen (Vanlauwe et al. , 

Despite these promising results, it should be kept in mind that only 
part of the light fraction or particulate organic matter consists of readily 
mineralizable materials, another part consists of recalcitrant materials, 
whose relative contribution increases with progressive decomposition of 
added plant materials in the soil (Palm et al., 2001; see Section 4.3). Also, 

Soil Nutrient Availability and Acidity 1 1 1 

in many cases the light or particulate organic matter fraction is too small 
to account for all the readily mineralizable nitrogen in the soil (e.g. 2-3% 
of total soil nitrogen in the study by Barrios et al., 1997) and it does not 
include labile pools such as microbial biomass which is mainly associated 
with clay-sized fractions (Palm et al, 2001). 

Chemical fractionation methods 

These difficulties could possibly be overcome by combining physical 
fractionation methods with chemical indices of nitrogen availability. An 
approach to quantifying a labile fraction of soil organic matter through 
selective oxidation with permanganate has been discussed in Section 4.4. 
Acid permanganate has also been used to extract ammonium-N from soil 
through selective oxidation of soil organic matter (Stanford and Smith, 
1978). The extracted amount was found to be closely correlated with 
potentially mineralizable nitrogen and thought to represent a fraction of 
soil nitrogen that is readily susceptible to biological mineralization. 
Gianello and Bremner (1986) found a close correlation between 
ammonium-N extracted by boiling the soil in 2 M KC1 for 4 h and 
potentially mineralizable nitrogen for 33 Brazilian soils of widely differing 
clay and organic matter contents. Egoumenides (1990) separated the total 
nitrogen in a large number of tropical soils by acid hydrolysis (6 N HC1 
for 16 h) into a non-hydrolysable fraction (mainly phenol and quinone- 
amino acid complexes), a hydrolysable-distillable fraction (mainly 
hexosamines, amides and certain amino acids) and a hydrolysable-non- 
distillable fraction (mainly amino acids of microbial origin). Greenhouse 
and field studies showed that nitrogen is supplied to plants principally 
from the hydrolysable-non-distillable fraction, which decreases under 
cropping and is sensitive to management measures such as crop rotations 
and fallowing. The hydrolysable-distillable fraction, in contrast, is mainly 
influenced by soil conditions, such as clay content. Numerous other 
chemical methods for the assessment of labile organic nitrogen have been 
described (Keeney, 1982). 

Microbial biomass nitrogen 

The microbial biomass is another labile pool of soil organic matter and 
organic nitrogen as discussed in Chapter 4. However, nitrogen in microbial 
biomass is only sometimes correlated with nitrogen mineralization in the 
soil. This is not surprising, as the microbial nitrogen pool is too small to 
account for the amount of nitrogen mineralized from soil (Palm et al., 
2001), and mineralization of nitrogen from other organic materials results 

112 C. Schroth etal. 

from microbial activity rather than its biomass (Jenkinson, 1988a; Groot 
and Houba, 1995). 

The high temporal variability of microbial biomass nitrogen requires 
attention when defining sampling plans, and repeated measurements 
during a cropping season may often be necessary to obtain a representative 
picture (Haggar et a!,., 1993; Mazzarino et al., 1993). Care is also needed 
in the selection of methods when attempting to relate microbial biomass 
nitrogen to nitrogen mineralization across different soils. In a comparison 
of different grassland soils, Hassink (1995) found that nitrogen 
mineralization in the soil increased with nitrogen in the total microbial 
biomass as measured with the fumigation-incubation method, but that the 
amount of nitrogen mineralized per unit of microbial biomass nitrogen 
was higher for sandy than for fine-textured soils. In contrast, the 
relationship between nitrogen mineralization and nitrogen in the 'active' 
microbial biomass as measured by substrate-induced respiration was 
independent of soil texture. This effect of soil texture, which was 
confirmed for other soils by Franzluebbers et al. (1996), was explained by 
differences in physical protection of soil organic matter in coarse- and fine- 
textured soils, and the second method was recommended for comparisons 
across different soil types. Detailed measurement procedures for microbial 
nitrogen are given in Anderson and Ingram (1993) and Voroney et al. 

5.3 Methods for Soil Phosphorus (J. Lehmann) 

Phosphorus availability is a limiting factor for plant production in many 
agricultural soils (Fairhurst et al. , 1 999). This is especially true in the highly 
weathered soils of the humid tropics. At the same time, the global 
availability of phosphorus fertilizers is limited and known reserves may be 
exhausted in about 100 years with the current growth of phosphorus usage 
(Stevenson and Cole, 1999). In regions of the world without a history of 
use of phosphorus fertilizers, phosphorus deficiency is very common 
(Wild, 1988). A large portion of applied fertilizer phosphorus may be fixed 
to iron and aluminium oxides and is then not available for plant uptake. 
These facts make sound phosphorus management an imperative, 
especially in situations where funds for fertilizer purchases are limited, as 
in tropical smallholder agriculture. Agroforestry techniques can help to 
overcome some of these constraints (Buresh, 1999). However, because of 
generally low phosphorus concentrations in mulch materials (see Chapter 
6), low atmospheric inputs (see Chapter 9) and low release by mineral 
weathering, adequate applications of phosphorus fertilizer are necessary 
in permanent agriculture to ensure economic and ecological sustainability 
(Buresh et al., 1997; Newman, 1997). One-time replenishment of the soil 

Soil Nutrient Availability and Acidity 113 

phosphorus capital, e.g. with phosphate rock, has been proposed as an 
option for soil fertility management in impoverished soils (von Uexkiill, 
1986; Buresh^a/., 1997). 

In plants, phosphorus serves as a structural element in nucleic acids 
and plays an important role in energy transfer and other enzyme processes 
(Marschner, 1995). Common diagnostic properties of phosphorus 
deficiency are a darker green leaf colour due to higher chlorophyll 
contents (often with red pigments from anthocyanins), reduced leaf 
extension and a higher root-to-shoot ratio, since root growth is much less 
affected by phosphorus deficiency than shoot growth (Wild, 1988; 
Marschner, 1995). A high phosphorus supply is needed for nodulation of 
legumes and hence phosphorus deficiency can also seriously reduce 
biological nitrogen fixation (Marschner, 1995). The minimum phosphorus 
concentration required for nodulation of soybean was about 0.5 (Jg P H in 
external solution (Marschner, 1995). For some trees suitable for 
agroforestry, however, no enhanced biological nitrogen fixation was noted 
with phosphorus applications on phosphorus-deficient soils (Reddel et al., 
1997; see also Section 13.1). 

In contrast to nitrogen, soil phosphorus is almost entirely derived from 
primary minerals, mainly apatite. Inorganic phosphorus is present in the 
soil solution as phosphate (H 2 P0 4 7HP0 4 2 ~). It adsorbs to clay mineral 
surfaces by ligand exchange and by specific bonding (bridging ligands 
which form two ligand exchanges) or precipitates with calcium and 
aluminium depending on the soil reaction (Mott, 1988). The pH for 
optimum phosphorus availability ranges from 5.5 to 6.5 when determined 
in water. 

The results of phosphorus analyses are greatly affected by the point 
of sampling, both vertically and horizontally. Inorganic phosphorus is very 
immobile in soil and small differences in sampling depth may yield 
completely different soil phosphorus contents. Vegetation effects (Tiessen 
et al., 1999) and application of organic or inorganic phosphorus sources 
to certain crops during previous cropping seasons (Woomer et al., 1998) 
cause uneven soil phosphorus distribution at a field and farm scale. This 
spatial variability creates challenges for scientific experimentation (Buresh, 

Standard extraction procedures 

Extraction procedures for plant-available soil phosphorus use bicarbonate 
(Olsen), acid ammonium fluoride (Bray 1 and 2), hydrochloric and 
sulphuric acid (Mehlich 1, 'double acid'), acetic acid and ammonium 
fluoride (Mehlich 2 and 3), ammonium carbonate and diethylenetriamine 
penta-acetic acid (DTPA) (Fixen and Grove, 1990), or anion exchange 

114 C. Schroth etal. 

resins (van Raij et al., 1986). Inorganic phosphorus in the extract is most 
commonly analysed with the molybdate-ascorbic acid (molybdene-blue) 
procedure of Murphy and Riley (1962). Inorganic phosphorus yields may 
be affected by soil pretreatment: various studies have shown that air-drying 
of soil samples may lead to considerably higher (14-184%) extractable 
inorganic phosphorus due to liberation of phosphorus from microbial 
biomass (Sparling et al., 1985; Srivastava, 1998). Summaries of soil test 
methods for plant-available phosphorus and their interpretation have been 
provided by Olsen and Sommers (1982), Novais and Smyth (1999), Fixen 
and Grove (1990) and several chapters in Carter (1993). 

Adsorption experiments 

In phosphorus-fixing soils, phosphorus fertilizer may be rapidly 
transformed into unavailable forms. The extent to which this happens 
depends on soil mineralogical properties (e.g. clay and especially allophane 
content and ferruginous nodules; Tiessen et al, 1991), but also on the 
history of phosphorus applications and plant uptake. Two adjacent fields 
may have the same available phosphorus contents measured with a 
standard procedure, but respond completely differently to mineral 
phosphorus applications because their adsorption sites for phosphorus are 
saturated to a different degree. Adsorption experiments are a means of 
predicting the magnitude of fixation from applied inorganic phosphorus. 
Soil is equilibrated with different solutions containing increasing 
phosphorus concentrations, and adsorption isotherms are determined 
from the relation between adsorbed and dissolved phosphorus (Fox and 
Kamprath, 1970). From the proportion of immobilized to added 
phosphorus, the fertilizer quantity that is needed to increase the 
concentration in solution to the desired level can be calculated (Bache and 
Williams, 1971; Holford, 1979; Haggar et al., 1991). By determining 
adsorption isotherms, it was shown that applications of organic matter 
decreased phosphorus adsorption due to competition for adsorption sites 
(Iyamuremye and Dick, 1996). Hue (1991) showed that organic acids 
decreased phosphorus sorption, but more so when phosphorus was added 
to the soil after the acids than when the phosphorus was added first. 
Accordingly, adsorption did not decrease if soluble phosphorus fertilizer 
(triple superphosphate) was added together with mulch on an Alfisol from 
East Africa (Nziguheba et al., 1998). The extent of the reduction of 
phosphorus sorption may depend not only on the type of organic acids 
(Hue, 1991), but also on the phosphorus content of the applied organic 
matter (Singh and Jones, 1976). Recent greenhouse studies have shown 
that high calcium contents of leaf mulch may decrease the solubility of 
phosphate rock. Consequently, phosphate rock did not increase maize 

Soil Nutrient Availability and Acidity 


growth when it was applied together with mulch (Smithson, 1999). In 
contrast, the dissolution of phosphate rock was increased by farmyard 
manure (Ikerra et al., 1994). 

Sequential fractionation methods 

A central goal of some agroforestry practices is to increase nutrient 
efficiency by keeping applied nutrients in available form or to make 
recalcitrant nutrients available (Sanchez, 1995). Application of organic 

0.5 g soil 

Resin strip, 16 h 

0.1 M NaOH, 16 h 

1 M HCI, 16 h 





Cone. HCI, 10 min, 80°C 


Residual P 

Resin P. 

Bicarbonate P, 
Bicarbonate P n 

Hydroxide P, 
Hydroxide P 

Dilute Acid P 

Cone. HCI Pi 
Cone. HCI P„ 

Fig. 5.3. Flow chart of the sequential phosphorus extraction method (after Tiessen 
and Moir, 1993). P : , inorganic phosphorus; P , organic phosphorus; cone, 

Table 5.1. Schematic comparison of plant availability and characteristics of soil phosphorus pools obtained by different analytical 

Plant availability 

Not available 

Slowly available 

Readily available 

Immediately available 

P pools 

Occluded inorganic 

Adsorbed inorganic 

Exchangeable inorganic 

Soluble inorganic 





long-term mineralizable 

medium-term mineralizable 

short-term mineralizable 

rapidly mineralizable 





Soil solution extraction 


Bray, Olsen, Mehlich 

(inorganic) 3 



Isotopic exchange 



Microbial biomass 

organic b 


Sequential extraction 





Physical fractionation 









a Brackets indicate less information about the respective pool from the applied method. 
b Only part of short-term mineralizable organic P. 

c No differentiation according to plant availability possible; generally only NaOH-extractable P. 
NMR, nuclear magnetic resonance. 

Soil Nutrient Availability and Acidity 117 

nutrient sources such as compost, manure or mulch and the 
transformation of recalcitrant nutrient pools into more readily available 
organic nutrient sources are key aspects of agroforestry. Organic nutrient 
management also implies that changes in available phosphorus will not be 
adequately assessed by conventional extraction methods for inorganic 
phosphorus. Mineralization and immobilization processes, and con- 
sequently the flow from inorganic to organic phosphorus pools and vice 
versa, become more important. Therefore, a single analysis of available 
phosphorus will very often not be sufficient to estimate the ability of a soil 
to supply phosphorus to plants. 

Sequential fractionation methods have been used for determining 
various soil organic and inorganic phosphorus pools with differing 
ecological properties. The Hedley fractionation method (Hedley et a!,., 
1982) has been widely adopted and latterly further modified (Fig. 5.3). 
The basic concept of the method is to sequentially extract several soil 
phosphorus compounds from the same soil sample by using extractants 
with increasing strength. The fractions that are extracted first have high 
plant availability, whereas the fractions that are extracted last have very 
low plant availability (Table 5.1). Organic phosphorus is determined by 
subtraction of inorganic phosphorus (determined after precipitation of 
organic matter) from total phosphorus (determined in digested samples; 
Tiessen and Moir, 1993). Bicarbonate and hydroxide extractions can also 
be carried out on separate subsamples to avoid phosphorus losses during 
the sequential soil treatment (Nziguheba et al., 1998; Mutuo et al., 1999). 
Soluble organic phosphorus is usually not determined by resin extraction, 
but may be an important pool and can be obtained from soil solution 
extraction (see Section 7.2). Bowman et al. (1998) suggested a more rapid 
method for analysing occluded and resistant soil phosphorus by 
subtracting acid-extractable phosphorus in an ignited sample from total 
soil phosphorus. A single-step method for determining organic soil 
phosphorus using concentrated sulphuric acid was found suitable for 
Nigerian savanna soils (Agbenin et al., 1999). 

With sequential fractionation it has been possible to assess the 
distribution of applied phosphorus in the soil profile (Beck and Sanchez, 
1996), the proportion of biological to geochemical phosphorus (Cross and 
Schlesinger, 1995), the soil incorporation of phosphorus from mulch or 
manure (Ikerra et al., 1994; Iyamuremye et al, 1996; Nziguheba et al., 
1998; Solomon and Lehmami, 2000), the effects of burning (Ball-Coelho 
et al, 1993) and tree effects on soil organic phosphorus (Tiessen et al, 1992, 
1999; Solomon and Lehmami, 2000; Lehmann et al., 2001c). 

Tree-specific phosphorus changes, especially in the dilute acid and 
organic phosphorus fractions, were found in a multistrata agroforestry 
system on an Oxisol in central Amazonia (Lehmann et al., 2001c), whereas 
rice monocropping, legume pasture and native savanna affected inorganic 

118 C. Schroth etal. 

phosphorus in the bicarbonate and hydroxide pools of a Colombian Oxisol 
(Friesen et al., 1997). The inorganic as well as organic bicarbonate and 
hydroxide phosphorus fractions did not show fertilizer effects as clearly as 
resin phosphorus in a tropical Alfisol (Mutuo et al., 1999). An equilibrium 
between soil fractions seems to exist which replenishes plant-available 
resin-P; from more recalcitrant phosphorus forms (bicarbonate and 
hydroxide) upon phosphorus uptake by plants (Schmidt et al., 1997). With 
phosphate rock applications, a combined anion-cation resin gives abetter 
measure of phosphorus availability than an anion resin alone, because it 
binds soluble calcium, thereby increasing phosphate solubility (Saggar et 
al., 1992). In shifting cultivation in north-eastern Brazil, decline of 
phosphorus fertility resulted from mineralization of organic phosphorus 
and subsequent fixation to the mineral soil matrix rather than from net 
phosphorus export (Tiessen et al., 1992). 

A problem with phosphorus fractionation is that the ecological 
functions of the different fractions are not clear-cut and may depend on 
soil properties and management. The first two steps of the Hedley 
fractionation, resin and bicarbonate extraction, are standard methods for 
plant-available phosphorus (see above). However, the more recalcitrant 
fractions can be difficult to interpret. For example, the exchange between 
occluded and adsorbed phosphorus (characterized by acid-extractable and 
residual phosphorus) on the one hand and plant-available phosphorus on 
the other hand depends on their strength of adsorption. Also, the 
phosphorus supply capacity from organic sources may not be adequately 
assessed with sequential fractionation, as the activity of acid or alkaline 
phosphatase in soil may be more sensitive. In soils with high mineralization 
capacity, Oberson et al. (1996) found higher acid phosphatase activity after 
application of organic materials than in the control, whereas bicarbonate 
P did not show any significant differences. 

A method for the direct measurement of phosphorus mineralization 
has recently been developed using phosphorus isotopes (Oehl etal., 2001), 
but has not yet been tested in tropical soils. 

Microbial biomass phosphorus 

Next to nitrogen, phosphorus is the most abundant nutrient in microbial 
biomass. It constitutes a highly active pool with large turnover rates and 
short-term mineralizability (Brookes et al, 1984) (Table 5.1). The analysis 
follows the fumigation procedure using bicarbonate extraction as 
described by Brookes et al. (1982). The extraction of phosphorus from 
fumigated soil with strips of anion exchange membranes was found to be 
more rapid and accurate than that with bicarbonate, especially in strongly 
phosphorus-fixing soils (Saggar et al., 1990; Kuono et al, 1995). The soil 

Soil Nutrient Availability and Acidity 119 

water content prior to the extraction has an important impact on microbial 
phosphorus results and needs to be monitored (Sparling and West, 1989). 
Microbial phosphorus was found to be the second most sensitive indicator 
of soil phosphorus differences between intensive fallow and maize 
monoculture after phosphorus in the light fraction of soil organic matter 
(Maroko et at., 1999). Effects of applications of Tithonia diversifolia mulch 
on microbial soil phosphorus were inconsistent between different studies 
from western Kenya. Small effects of organic phosphorus applications 
may not be readily detected in soil microbial phosphorus (Jama et a!,., 


Adsorption experiments can also be conducted with 32 P to determine 
isotopically exchangeable phosphorus (Fardeau et ai, 1996; Frossard and 
Sinaj, 1997). Whereas green manure did not affect phosphorus extracted 
by Bray's solutions or resin or phosphorus adsorption on an Oxisol in 
Brazil, it increased the amount of isotopically exchangeable phosphorus 
(LeMare et ai, 1987). Exchange processes between organic and inorganic 
soil phosphorus pools have also been studied with the radioisotopes 32 P 
and 33 P (Di et ai, 1997). The difference between exchangeable 32 P on a 
non-incubated soil (only physicochemical exchange) and the 32 P measured 
from incubated soils labelled with 32 P (physicochemical exchange plus 
phosphorus mineralization) gives an estimate of phosphorus 
mineralization (Lopez-Hernandez et ai, 1998). Isotopically labelled 
phosphorus fertilizer can also be used to follow its fate in soil phosphorus 
fractions (He and Zhu, 1997) or in the soil profile as affected by organic 
applications (Othieno, 1973). Additionally, the utilization of different soil 
phosphorus pools by different plant species may be estimated by 32 P 
labelling of the soil: plants utilizing unlabelled phosphorus are able to 
access phosphorus pools that are more recalcitrant than the isotopically 
exchangeable, soluble phosphorus fraction as shown for lupin (Braum and 
Helmke, 1995). For valid interpretation of radioisotope experiments, care 
has to be taken to meet underlying assumptions of isotope exchange 
kinetics (Frossard and Sinaj, 1997). 

Other methods 

From the total organic phosphorus in soil, no more than one-third has 
been chemically identified (Stevenson and Cole, 1999). Specific 
phosphorus compounds in soil comprise inositol phosphates, 
phospholipids, nucleic acids, phosphoproteins and metabolic phosphates 

120 C. Schroth etal. 

such as ATP. Organic and inorganic bonding of soil phosphorus can be 
analysed by nuclear magnetic resonance ( 31 P NMR) spectroscopy (Newman 
and Tate, 1980). With this method, increased amounts of labile organic 
phosphorus and, following mineralization, also of labile inorganic 
phosphorus were found after grassland conversion into conifer plantations 
(Condron et al., 1996). Relatively labile diester-P compounds were shown 
to accumulate in soil from earthworm casts (Guggenberger et al., 1996) 
and following application of manure (Solomon and Lehmann, 2000). 

A combination of physical fractionation of soil (see Section 4.3) with 
phosphorus analysis can increase options for assessing transformation 
processes which are not detected by analyses of phosphorus in the bulk 
soil. This approach may, in many cases, be more useful than analysing 
organic phosphorus fractions on bulk soils, because it provides information 
on turnover and mineralizability of soil organic matter and therefore also 
of associated organic phosphorus (Tiessen et al., 1999). Total phosphorus 
analyses, sequential phosphorus fractionation and 31 P NMR spectroscopy 
have been used for phosphorus determination in density and particle size 
separates (Maroko et al., 1999; Solomon and Lehmann, 2000). Maroko et 
al. (1999) found that phosphorus effects of planted Sesbania sesban fallows 
and natural fallows, in comparison with continuous maize cropping, were 
most evident in the light organic matter fraction compared with a range 
of other phosphorus analyses. 

Soil phosphorus availability for plants is influenced by the presence 
and activity of phosphorus-solubilizing bacteria in the rhizosphere, root 
release of protons and organic acids, mycorrhizal root infection, and root 
hair length and density. These factors are more important for the uptake 
of phosphorus than for that of any other macronutrient due to the low 
mobility of phosphorus in soil and its strong fixation to oxides. The 
ultimate test for soil phosphorus availability is plant uptake. Bioassays can 
give valuable information about the effects of different management 
interventions such as applications of mulch or mineral fertilizer on plant 
phosphorus uptake (Othieno, 1973; Haggar et al. , 1991; Jama et al., 2000). 

Phosphorus transfers in soil, e.g. from topsoil to subsoil, are seldom 
determined due to experimental problems of obtaining phosphorus from 
soil solution with ceramic suction cups or lysimeters (see Section 7.2). With 
little fertilization, amounts of inorganic phosphorus are usually low and 
possess low mobility especially in high phosphorus-fixing soil. In the 
absence of soil erosion, phosphate losses from agricultural soils do not 
usually exceed 0.5 kg ha -1 year 1 (Sharpley and Withers, 1994). However, 
organic phosphorus can be the dominant phosphorus compound in soil 
solution and dissolved organic carbon properties largely determine 
phosphorus mobility in soil (Quails et al. , 1991; Donald etal., 1993). Donald 
et al. (1993) demonstrated for a forest soil that 64% of the organic 
phosphorus was contained in the hydrophobic neutral fraction of dissolved 

Soil Nutrient Availability and Acidity 121 

organic carbon that was only weakly adsorbed to the soil. Therefore, soil 
solution phosphorus can be mobile in its organic form (Schoenau and 
Bettany, 1987) and should be considered when calculating phosphorus 

5.4 Methods for Soil Sulphur (J. Lehmann) 

Arable soils of the humid tropics often have low total sulphur contents 
because of low contents of parent materials, strong weathering and high 
leaching losses. In comparison with other macronutrients, sulphur received 
little attention as a plant nutrient in tropical crop production in the past 
and few results have been published (Kang et al. , 1 98 1 ; Acquaye and Kang, 
1987; Motavalli et al. , 1993). Sulphur deficiency in tropical agroecosystems 
has recently increased, however, due to the more common use of nitrogen 
and phosphorus fertilizers with low sulphur contents (e.g. substitution of 
ammonium sulphate by urea and of single by triple superphosphate) as 
well as higher crop yields in some regions (Ceccotti et al., 1998). 

Up to 90% of sulphur in plants is present as amino acids and therefore 
bound in proteins. Sulphur is an essential element for many functions in 
plants (e.g. synthesis of chlorophyll) and sulphur deficiency not only 
decreases plant growth but also forage quality (Tisdale, 1977) and 
resistance to pests and diseases. 

Visual diagnosis of sulphur deficiency in plants is difficult and criteria 
range from chlorosis in young leaves, starting from the edges, spoonlike 
deformations, succulence, poor pod formation, to stunted growth (Schnug 
and Haneklaus, 1998). Deficiency symptoms are more difficult to diagnose 
for monocotyledons than for dicotyledons and resemble those of nitrogen. 
Tissue analysis of total sulphur is a more reliable indicator of sulphur 
deficiency than visual diagnosis if the usual sampling criteria are followed, 
such as the collection of defined plant organs and development stages 
(Schnug and Haneklaus, 1998). 


Plants take up sulphur in the form of S0 4 2 _ , which is the most common 
form of inorganic sulphur in agricultural soils. Grasses have a higher ability 
than legumes to utilize sulphate in soils (Havlin et al., 1999), and excessive 
competition for sulphate-S in legume-non-legume mixtures may cause 
decreases in biological nitrogen fixation. Common extractants for plant- 
available sulphate-S include water, calcium chloride, sodium phosphate 
(Na 2 HP0 4 ) (Kowalenko, 1993a), Na-acetate/acetic acid (Landon, 1991), 
lithium chloride (Tabatabai, 1982) and exchange resins (Searle, 1992). 

122 C. Schroth etal. 

However, the value of analysing sulphate-S for estimating the availability 
of soil sulphur for plants is not entirely clear (Schnug and Haneklaus, 
1998) and additional availability tests are recommended. 

Sulphate sorption characteristics are important for quantifying 
sulphur leaching and retention in tropical soils. Their determination is 
analogous to phosphorus sorption experiments (see Section 5.3). In soils 
high in clay, organic matter and exchangeable acidity, soil drying prior to 
analysis can result in large errors and experimentation with field moist 
samples is recommended (Comfort et al., 1991). The extent of sulphate 
sorption in acid subsoils can differ from that of the topsoils as shown for 
an Alfisol and Oxisol from India (Patil et al., 1989), but recycling by deep- 
rooting trees as occurs with nitrate (Box 8.1) has not yet been the subject 
of research. Adsorption generally increases with higher clay contents, 
oxide contents, and lower pH. Adsorption and precipitation of sulphate- 
S by carbonates has to be considered in certain dryland soils. 

Mineralization of sulphur from soil organic matter and biomass 

More than 95% of sulphur in soil occurs in an organic form. Since organic 
sulphur cannot be determined directly, total sulphur is usually measured 
using dry combustion and gas chromatography or dry ashing with alkaline 
oxidation followed by ion chromatography (Tabatabai et al. , 1 988). Organic 
sulphur is calculated as the difference between total and inorganic sulphur. 
Sulphur availability for plants is closely linked to the mineralization of 
organic sulphur, similar to soil nitrogen (see Section 5.2). Sulphur 
mineralization can be measured by incubation methods. However, much 
higher mineralization rates of sulphur (but not nitrogen) have been 
measured when the mineralized sulphur was periodically leached out of 
the soil than in closed systems, indicating that certain incubation methods 
that are commonly used for nitrogen would lead to underestimation of 
sulphur mineralization rates (Maynard et al., 1983). Therefore, mineraliz- 
ation studies are best conducted in open systems under field conditions 
(Eriksen et al., 1998), where leaching of sulphate-S is allowed and mineral- 
ization processes are controlled by ambient temperature and moisture 
conditions. Plants have been shown to stimulate sulphur mineralization in 
the soil through a greater proliferation of microorganisms in the 
rhizosphere (Stevenson and Cole, 1999), which can make the prediction 
of sulphur availability more difficult when estimated from mineralization 
studies where roots are excluded. 

The effect of sulphur from organic sources such as leaf mulch, animal 
manure or compost on sulphur availability depends on the concentration 
of sulphur in the organic materials. Sulphur immobilization in microbial 
biomass occurred after application of barley straw with low sulphur content 

Soil Nutrient Availability and Acidity 123 

(2.2 mg g- 1 ), but not of rape leaves with high sulphur content (7.2 mg g- 1 ; 
Wu et al., 1993). However, only a rough threshold value can be given for 
sulphur release from organic residues, which tend to mobilize sulphur if 
the C:S ratio is <200 and immobilizes it with a C:S ratio >400 (Stevenson 
and Cole, 1999). Sulphur release was found to be quite different from that 
of nitrogen, as soluble sulphate-S is initially solubilized in high amounts 
and sulphur mineralization is not closely related to the C:S ratio of the 
organic material (Janzen and Ellert, 1998). Short-term increases of 
available sulphur from applied animal manure can rarely be expected, 
although long-term accumulation of manure or urine in pasture soils 
increases soil organic sulphur (Saggar et al, 1998) and leads to a higher 
supply of sulphate-S. 

Fractionation of soil organic sulphur 

Another approach to determining soil organic sulphur pools with different 
mineralizability is chemical fractionation. Ester-sulphate-S compounds can 
be analysed using reduction with hydriodic acid (HI) and subtracting 
inorganic sulphate-S. Carbon-bonded sulphur is then obtained by 
subtraction of ester-S from organic sulphur (Fig. 5.4). Hi-reducible ester- 
sulphate-S is considered to be more labile than carbon-bonded sulphur, 
although some authors have reported the opposite (Janzen and Ellert, 
1998). Carbon-bonded sulphur was not affected by trees in a multistrata 
agroforestry system in the central Amazon, whereas ester-sulphate-S 
increased in soil underneath trees with large nutrient recycling (Lehmann 
etal., 2001c). Solomon et al. (2001) reported that forest clearing and maize 
cultivation in the Ethiopian highlands reduced both the carbon-bonded 
and ester-sulphate-S, whereas, in plantations of tea and Cupressus lusitanica, 
only the carbon-bonded sulphur fraction decreased. In a simplified 
conceptual model, carbon-bonded sulphur is considered to be released by 
mineralization, which is controlled by the needs of soil microorganisms for 
organic carbon, whereas ester-sulphate-S is released by enzymatic 
hydrolysis and therefore controlled by the supply of sulphur (McGill and 
Cole, 1981; see also comments in Ghani et al., 1992). This dichotomous 
cycling of sulphur also explains variable N:S ratios of mineralized organic 
matter. Usually, sulphur was found to be more stable in organic matter 
than nitrogen upon soil cultivation (Stevenson and Cole, 1999). The 
dynamics of sulphur fractions under long-term cultivation have also been 
studied in humin, humic and fulvic acids (Bettany et al., 1980); however, 
the analysis of ester-sulphate-S and carbon-bonded sulphur in these 
extracts may be affected by hydrolysis of organic sulphur compounds 
during the extraction and may not yield reliable results (Freney, 1986). 
The analysis of organic sulphur in soil physical fractions offers the 


C. Schroth ef al. 

0.5 g soil 

0.01 M CaCI 2 , 30 min 
drying at 100"C 

0.5 g soil 

0.5 g soil 

Na-hypobromite, 3 min 
drying at 250'C 

Distillation with HI acid 

Distillation with HI acid 

Distillation with HI acid 

Inorganic sulphate-S 

Total Hl-reducible S 

Total S 


Total S - Inorganic sulphate-S = Organic S 

Total Hl-reducible S - Inorganic sulphate-S = Ester-sulphate-S 

Organic S - Ester-sulphate-S = Carbon-bonded S 

Fig. 5.4. Conceptual outline of the sulphur fractionation according to Kowalenko 
(1993b). Total sulphur can also be obtained by dry combustion with an automated 

possibility to detect tree effects on soil sulphur more sensitively and also 
to evaluate its stability after applications of mulch or animal manure. 
Sulphur can be analysed after fractionation according to particle or 
aggregate size and density (Anderson et al. , 198 1). A fractionation method 
has been described which uses acetylacetone as extractant and different 
intensities of ultrasonic dispersion (Eriksen et al., 1995). 

Microbial sulphur comprises a pool of highly reactive sulphur in soils 
and is important for understanding the mineralization of carbon-bonded 
sulphur, although it only amounts to 1-2.5% of total soil sulphur. Microbial 
sulphur can be determined by the fumigation-extraction method using 
calcium chloride extraction (Saggar et al., 1981; Wu et al., 1994). Other soil 
organic sulphur measurements such as analyses of enzymes, amino acids 
and sulpholipids have been developed, but have not yet yielded easily 
interpretable results. 

Other methods 

Isotopic techniques using radioactive 35 S have been successfully employed 
to identify the fate of added sulphur in different soil pools and its 
subsequent mineralization (Freneyetal., 1971; Maynard et al. , 1983; Ghani 
et al., 1993). Soil is incubated with carrier-free 35 S0 4 2 ~ in the presence or 

Soil Nutrient Availability and Acidity 125 

absence of carbon sources (glucose) and sulphate, and analysed for 35 S. 
Excess sulphate is leached and the soil is re-incubated. With this method, 
processes such as the incorporation of added sulphur from mulch or 
animal manure and the incorporation into soil sulphur pools such as ester- 
sulphate-S, carbon-bonded sulphur and sulphate-S can be studied. 

When studying the sulphur balance of agricultural systems, exchanges 
of sulphur between the vegetation and the atmosphere have to be 
considered (Krouse et al., 1991). If atmospheric sulphur concentrations 
are high, such as near the sea, sulphur uptake by plant leaves from the air 
can sometimes occur in sufficient amounts to meet plant requirements. 
On the other hand, plants may also release volatile sulphur compounds 
in response to nutritional or light stress. Sulphur losses from decomposing 
litter can be greater than from living plants (Janzen and Ellert, 1998). 

5.5 Methods for Potassium, Calcium and Magnesium in Soil 

(G. Schroth, J. Lehmann) 

The cations potassium, calcium and magnesium occur in several forms in 
the soil that differ in their availability to plants. The most readily plant- 
available fraction is that in the soil solution, followed by the exchangeable 
fraction, which replenishes the soil solution if nutrients are removed by 
either plant uptake or leaching. Potassium fixed in clay interlayers becomes 
available at a time scale from hours to weeks. The least available forms are 
various primary and secondary soil minerals, which release the respective 
nutrients upon weathering (Haby et al, 1990). Together with sodium 
(where this is present at relevant quantities), potassium, calcium and 
magnesium make up the exchangeable bases of a soil. Numerous soil test 
methods are used for assessing the availability of these nutrients to plants, 
which are discussed in detail by Haby et al. (1990). As a rule, those methods 
that have been found most useful in the respective study region and for 
which locally calibrated reference values exist should be used. Of particular 
interest are multielement extractants such as Mehlich 3, which allows the 
simultaneous extraction of the exchangeable bases, phosphorus and 
various micronutrients (Tran and Simard, 1993). 

Acid tropical soils, such as Oxisols and Ultisols, are characterized by 
low base saturation, and correspondingly low exchangeable contents of 
potassium, calcium and magnesium. Despite low contents, potassium 
availability rarely limits crop yields when acid soils are first taken into 
cultivation, although it becomes an important factor for permanent, 
intensive cropping once these soils have been limed and other deficiencies 
(e.g. phosphorus) have been corrected (von Uexkull, 1986). Small 
quantities of lime often increase potassium uptake by plants due to 
improved root growth, whereas larger quantities may depress potassium 

126 C. Schroth etal. 

uptake by increasing the cation exchange capacity associated with variable 
(pH-dependent) charges in the soil, thereby reducing potassium 
concentrations in the soil solution, and by antagonistic effects of calcium 
ions on potassium uptake. Low magnesium contents are common in acid 
soils, and magnesium deficiency can be induced by high aluminium 
contents (see Section 5.6) or by potassium fertilization. Tree crops tend to 
be more sensitive to magnesium deficiency than annual crops, and annual 
dicotyledons more sensitive than annual monocotyledons (von Uexkiill, 

With time under cropping the exchangeable soil contents especially 
of calcium and magnesium decrease and exchangeable acidity increases, 
unless cation losses caused by the exportation of harvested products, 
leaching and erosion are replaced by corresponding inputs of fertilizers 
and lime (Pieri, 1989; Juo et al., 1995; Smyth and Cassel, 1995). In most 
soils, calcium and magnesium are more susceptible to leaching than 
potassium (see Section 7.1). If land is left to regenerate under spontaneous 
or planted fallow vegetation, losses of calcium (and probably other cations) 
in the plant-soil system may continue during the first years before 
nutrients start to accumulate due to atmospheric inputs, mineral 
weathering (where weatherable minerals are still present) and uptake from 
the subsoil (Szott et al, 1999). 

Tree species differ in their effects on nutrient cations in the soil, and 
such species -related differences can even be detected within closed forest 
stands (Finzi et al., 1998). Trees are expected to reduce nutrient leaching 
from cropping systems (see Chapter 7), but they may also immobilize 
relevant quantities of nutrients in their biomass. In a long-term 
experiment, Hulugalle (1994) found lower exchangeable calcium contents 
in the soil under hedgerow intercropping than annual crops and pasture 
and explained this in terms of high calcium demand of the trees. Larger 
amounts of calcium and magnesium were immobilized in unpruned Cordia 
alliodora than in pruned Erythrina poeppigiana shade trees in coffee 
plantations (Beer et al., 1998), and immobilization of potassium in the 
Cordia alliodora biomass was suspected to be a potential limiting factor to 
crop and tree productivity (Beer, 1988). 

Some trees promote the accumulation of cations in the topsoil, 
presumably through efficient uptake by a large root system and subsequent 
release from litter. Reported cases include calcium accumulation by 
Gmelina arborea (Sanchez et al., 1985; Fisher, 1995) and Senna siamea 
(Drechsel et al, 1991). Elevated calcium contents also characterize the leaf 
litter of Terminalia superba (F. Bernhard-Reversat, unpublished data). 
Several tree species increased the calcium and potassium contents in the 
topsoil of a degraded rainforest site (Fisher, 1995). Relatively high 
potassium concentrations are found in the biomass of the large herbaceous 
plant, tithonia (Tithonia diver sifolia), which is also rich in nitrogen and 

Soil Nutrient Availability and Acidity 127 

phosphorus. The effectiveness with which this potassium was used by 
maize when supplied with tithonia biomass seemed to be comparable to 
the use of potassium from mineral fertilizer (Jama et al., 2000). Increased 
potassium contents in the topsoil under certain cover crops have been 
explained with potassium redistribution from lower soil horizons (Smyth 
etal., 1991; Barber and Navarro, 1994). 

In accord with their different functions in the plant, potassium and 
calcium differ strongly in their release dynamics from both dead and living 
biomass. Potassium is rapidly leached from leaf and woody litter, whereas 
calcium often shows an absolute increase in biomass during the first stages 
of decomposition, possibly due to a combination of slow release and 
calcium import in fungal hyphae (Swift et al., 1981; Maheswaran and 
Gunatilleke, 1988; Schroth et al., 1992). Potassium is also leached in 
relatively large quantities from living plants. In studies in a rainforest in 
Borneo (Burghouts et al., 1998) and in multistrata agroforestry and fallow 
in Amazonia (Schroth et al. , 200 la), potassium was the only macronutrient 
for which larger quantities were recycled from the standing biomass to the 
soil in throughfall and stemflow than in litterfall. 

Cycling of calcium, magnesium and potassium in agroforestry systems 
can also be studied with tracers. Due to handling difficulties in conjunction 
with health hazards, radioisotopes ( 42 K, 28 Mg, 45 Ca) are only used in 
specialized experimentation under controlled conditions. However, 
subsurface applications of the elements lithium and rubidium (for 
potassium) and strontium (for calcium) can be used to assess root activity 
distribution (Fitter, 1986; van Rees and Comerford, 1986) (see Section 
8.2). The determination of the fate of applied potassium and calcium 
fertilizer with these tracers may pose difficulties, because they would have 
to be applied at high concentrations, which would alter their behaviour in 
soil and during plant uptake. 

5.6 Methods for Soil Acidity (C. Schroth) 

Acid soils, especially Oxisols and Ultisols, occupy vast areas in the tropics. 
Naturally, these soils are mostly covered by forest or savanna, but during 
the last decades increasing areas have been cleared for agriculture. Despite 
the acidity and nutrient deficiency of these soils, their economic 
importance is substantial: 100% of the world tea, rubber, oil palm and 
cassava production and 90% of the world coffee production come from 
acid soils (von Uexkull and Mutert, 1995). However, the conversion of 
forest on acid soils for agricultural use very often leads to the development 
of unproductive, man-made savannas, such as the imperata grasslands of 
South-east Asia and some abandoned pastures of Amazonia. In the absence 
of the necessary inputs of nutrients and lime, the residual fertility from 

128 C. Schroth etal. 

the clearing of the vegetation is lost after only a few crop harvests, and the 
land is then commonly abandoned. Under conditions of increasing 
pressure on acid soils in the tropics from a growing population, the 
problems of soil acidity and infertility are intimately connected to losses in 
tropical forest area and biodiversity (von Uexkiill and Mutert, 1995). 

Acid soils require a humid climate for their development and are thus 
typically found in tropical savanna and rainforest zones. Under 
agricultural use, soil acidification is increased by leaching of basic cations 
and their removal with harvested crops; acidifying fertilizers such as urea, 
ammonium sulphate and potassium chloride; and the use of nitrogen- 
fixing legumes, for example, in pastures (Rowell, 1988; von Uexkiill and 
Mutert, 1995). Agroforestry techniques may reduce problems of soil acidity 
by reducing leaching losses (see Chapter 7) and by increasing the level of 
soil organic matter, which detoxifies dissolved aluminium (see Chapter 4), 
although increased acidification due to the use of nitrogen-fixing trees 
may also be expected (van Miegroet and Cole, 1984). 

Acid soils present a number of interrelated problems, including 
toxicity of aluminium, manganese and (under reducing conditions) iron 
as well as deficiencies of phosphorus, calcium, magnesium, potassium and 
micronutrients (e.g. molybdenum, boron). Low water-holding capacity 
(e.g. of Oxisols) and susceptibility to crusting, erosion and especially 
compaction (Oxisols, Ultisols) aggravate the problem (Marschner, 1995; 
von Uexkiill and Mutert, 1995). Here, only aluminium toxicity as a key 
factor of acidity stress in many mineral soils is discussed. The risk of 
manganese toxicity is lower than that of aluminium toxicity in highly 
weathered tropical soils, because these often have low total manganese 
contents (Marschner, 1995). Mechanisms of plant adaptation to soil acidity 
have been discussed by Marschner (1995) and, specifically for tree species, 
by Fisher and Juo (1995). Lists of acid-tolerant and intolerant annual and 
perennial crop and pasture species and varieties are provided by Sanchez 
and Salinas (1981), and much useful information on tree species for acid 
soils can be found in the volume edited by Evans and Szott (1995). 

Diagnosis of aluminium toxicity in plants 

In soils with a pH in H 2 >5.5, aluminium is tightly held to the exchange 
surfaces, but as the pH falls below 5, exchangeable aluminium and 
aluminium concentrations in the soil solution increase markedly and can 
cause aluminium toxicity in sensitive plant species (von Uexkiill, 1986). 
Neither visual diagnosis of shoots nor foliar analysis for aluminium is a 
suitable indicator of aluminium stress in cultivated plant species. Most of 
these are aluminium excluders and do not readily transport aluminium 
from the roots into the shoots, a notable exception being tea (Camellia 

Soil Nutrient Availability and Acidity 129 

sinensis). In fact, aluminium concentrations in shoots of healthy plants may 
be higher than in those of aluminium-stressed plants. In tropical 
rainforests, aluminium includers and excluders exist at the same site, and 
these may vary in their foliar aluminium contents by two orders of 
magnitude. Aluminium concentrations in roots of aluminium-stressed 
plants are correlated with the severity of damage, but the analysis of roots 
for aluminium is difficult due to adhering mineral soil (Bergmann, 1988; 
Marschner, 1995). 

Symptoms of aluminium toxicity first appear on the roots, which are 
shortened, thickened, show reduced branching and are brown to black in 
colour with black tips. Symptoms are limited to the actively growing tissue 
(Bergmann, 1988). Reduced root development, especially in the subsoil, 
makes plants sensitive to drought stress and reduces the access to nutrients 
in the subsoil, thereby also increasing nutrient leaching (Rowell, 1988). 
High aluminium concentrations in the solution impede the uptake of 
polyvalent cations such as calcium and magnesium and may thus induce 
deficiencies of these elements, whereas potassium uptake is usually 
unaffected (Marschner, 1995). Aluminium interacts with phosphorus 
availability through the reduction of root growth and by binding 
phosphate on root surfaces, cell walls and in the free space of plant roots. 
At higher aluminium concentrations, symptoms of aluminium toxicity 
may, therefore, be similar to those of phosphorus or calcium deficiency 
(Bergmann, 1988; Rowell, 1988). 

Aluminium measurement in soil 

To relate problems of aluminium toxicity and induced nutrient 
deficiencies in plants to soil properties, aluminium may either be measured 
in the dry soil or in the soil solution. The former approach is more 
compatible with standard soil sampling and laboratory practices and is, 
therefore, more common. Exchangeable acidity is usually measured by 
leaching a soil sample with 1 M KC1 and quantification of the extracted 
acidity (aluminium and protons) by titration (Hendershot et al., 1993b). 
Exchangeable acidity is given in mmol c kg-' and/or as a percentage of the 
effective cation exchange capacity measured at the pH of the soil, e.g. by 
summing the exchangeable cations (Hendershot et al., 1993a). In relatively 
aluminium-tolerant crop species and varieties, toxicity symptoms may 
appear at 60% aluminium saturation, and in less tolerant species and 
varieties at 30% aluminium saturation (von Uexkull, 1986). A procedure 
for fractionating aluminium forms in soils has been proposed by Soon 

However, plants are directly influenced by aluminium in the soil 
solution, which is not always proportional to exchangeable aluminium. In 

130 C. Schroth etal. 

a comparison of the soil and soil solution chemistry of different land-use 
systems in central Amazonia, fallow and low-input agroforestry had more 
exchangeable acidity than high-input agroforestry and monoculture 
plantations, but the high-input systems had more aluminium in the soil 
solution due to the increased availability of fertilizer cations for exchange 
reactions with the sorbed aluminium and the acidifying effect of 
nitrification of fertilizer nitrogen. Peak concentrations of aluminium in the 
soil solution coincided with periods of highest nutrient availability 
following fertilizer applications and could have negatively affected the 
uptake of fertilizer nutrients (Schroth et al., 2000b). Critical aluminium 
concentrations in the soil solution which negatively affect plant growth 
vary from about 1 mM for tolerant species to 1 u.M for sensitive species 
(Rowell, 1988). In legumes, the critical concentration for nodulation may 
be substantially lower than that for the host plant (Ashwath et al., 1995; 
Marschner, 1995). 

The principal difficulty in interpreting aluminium concentrations in 
soil solution (but also exchangeable aluminium levels) is that, depending 
on the solution composition, different aluminium forms coexist which may 
differ widely in their phytotoxicity (Kinraide, 1991). Organic substances 
such as fulvic acids, humic acids and other organic acids may detoxify 
aluminium by forming complexes. Consequently, aluminium in the subsoil 
can be phytotoxic at lower concentrations in the soil solution than in the 
topsoil where organic matter contents are higher (Marschner, 1995), and 
the application of mulch and green manure can reduce aluminium toxicity 
in acid soils (Harper et al, 1995). The concentration of different aluminium 
species in the soil solution can be predicted with models such as geochem, 
although not all the required stability constants may be accurately known 
(Kerven et al., 1995). Also, the actual solution composition can differ from 
that predicted by equilibrium models during peak flow conditions after 
heavy rain when aluminium release may be influenced by diffusion 
processes (Franken et al, 1995). Because of the suppressive effect of high 
aluminium concentrations on calcium and magnesium uptake, the molar 
Ca:Al and MgAl ratios in soil solution have been used as measures of 
aluminium-induced deficiencies of these elements and were superior in this 
respect to the concentrations of the individual elements (Marschner, 1995). 

In addition to the effects of soil acidity on plants via toxicity and 
induced deficiency phenomena discussed above, soil acidity may also 
influence plant growth by affecting soil biological properties such as the 
abundance and composition of the soil fauna (Lavelle et al, 1995) and plant 

Methods of determining lime requirements for crops of differing 
sensitivity to soil acidity have been discussed by Sumner (1997) and van 
Raij and Quaggio (1997). For the assessment of plant litter effects on soil 
acidity, see Chapter 6. 

Chapter 6 

Decomposition and Nutrient Supply 

from Biomass 


Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil 

6.1 Synopsis 

One of the benefits that may be expected from agroforestry in comparison 
with annual and perennial monocultures is an increased production of 
biomass. When returned to the soil, this biomass protects the soil surface, 
releases nutrients, replenishes soil organic matter and provides carbon 
substrates for soil biota. This chapter focuses on above-ground biomass, 
which is easier to manage and which has been studied more intensively 
than root biomass in agroforestry, but the principles are also valid for the 
latter. Methods for studying root turnover as an important source of 
biomass input in the soil are discussed in Section 12.4. 

With respect to nutrient inputs into the soil, there is a major difference 
between biomass grown in situ and fertilizers in so far as only part of the 
nutrients released from the biomass will be an external input, the 
remainder has been taken up from the same soil to which it is returned. 
Biomass transfer systems represent an intermediate situation in which the 
nutrients in the biomass are added to the site where the biomass is applied, 
but are removed from another site within the same landscape. Net nutrient 
additions in biomass to a system include biologically fixed nitrogen, 
nutrients taken up by the trees either from deeper subsoil horizons or from 
recalcitrant pools that would not have been readily available to crops. 
Nutrients taken up by extensive lateral tree roots beyond field boundaries 
are again an intermediate case. They can be seen as net additions to the 
system if they are taken up from a site that would have never been cropped 
(a river bed), but only as redistribution if they are taken up from a 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 131 

F.L. Sinclair) 

132 C.Schroth 

neighbouring field or from a fallow that will be cropped the following year, 
depending on the spatial and temporal scale under consideration. Other 
nutrients in biomass may not be additions, but rather 'avoided losses', such 
as nutrients taken up from the percolating soil solution that would have 
been leached in the absence of the trees (see Chapter 7). 

However, a significant proportion of the nutrients in tree biomass may 
have been taken up in direct competition with crops. If tree biomass is 
applied to the soil, these nutrients are merely recycled. In simultaneous 
agroforestry systems, this may often be the main part of the nutrients 
contained in tree biomass. In contrast, in fallows, where direct tree-crop 
competition for nutrients does not occur or only between neighbouring 
crop and fallow plots, much of the nutrients in the tree biomass may consist 
of net additions and avoided losses. 

The basic function of fallows is to produce biomass and thereby 
regenerate soil fertility. In the most widespread simultaneous agroforestry 
systems, the production of nutrient- rich biomass may also be an important 
function of the trees (e.g. pruned legume trees as shade in coffee and cocoa 
plantations), but it is usually not their only function and may often be of 
secondary importance to farmers in comparison with direct production 
functions (e.g. fruit or fodder trees in parkland systems in savannas). In 
hedgerow intercropping, the trees serve principally for the production of 
biomass and the recycling of nutrients through this biomass, but do not have 
direct production functions unless they are simultaneously used for fodder. 
The fact that this technique has not been widely adopted may indicate that 
these motives alone often do not provide sufficient incentive for farmers to 
plant and manage trees (Fujisaka, 1993). Agroforestry interventions need 
to be designed in the context of the niche within a farming system that the 
trees will fill. The optimization of carbon and nutrient supply will generally 
be constrained by farmer requirements for economic product, low 
competitiveness with crops and ease of management. 

Nutrient cycling in biomass in agroforestry systems 

The quantities of tree litter or primings that are produced in agroforestry 
systems and the amounts of nutrients therein can be substantial. In coffee 
and cocoa plantations shaded by legume trees in Central America, 3-14 
Mg ha _1 year 1 of primings are produced by the trees, containing 60-340 
kg of nitrogen (Beer et a!,., 1998). Leguminous trees in hedgerow 
intercropping produced up to 20 Mg ha _1 year 1 of prunings, containing 
as much as 358 kg of nitrogen, 28 kg of phosphorus, 232 kg of potassium, 
144 kg of calcium and 60 kg of magnesium (Palm, 1995). Very little is 
known about the amounts of carbon and nutrients released by root systems 
in agroforestry (see Section 12.4). Relevant nutrient release from roots 

Decomposition and Nutrient Supply from Biomass 133 

may occur when large amounts of superficial tree roots are destroyed by 
tillage at the beginning of a cropping season, and following the conversion 
of forest or fallow plots into crop fields. 

The amount of biomass produced by agroforestry trees depends on 
the tree species, the number of trees per hectare, tree age, tree 
management and site factors. The nutrient concentrations in different 
parts of the biomass depend mainly on tree species, phenological stage 
(e.g. senescence, fruiting), management and site factors. Nitrogen-fixing 
trees normally have higher nitrogen concentrations in the biomass than 
non-fixing species, but this characteristic also varies widely between species 
(Palm, 1995). Deciduous species generally have higher nitrogen 
concentrations in the leaves than evergreen species (Eamus, 1999). Certain 
monocots such as bananas, bamboos and palms have relatively high 
potassium concentrations in their biomass (Sanchez et al, 1985). Tree 
prunings normally have higher concentrations of mobile nutrients such 
as nitrogen, phosphorus, potassium and zinc than naturally fallen litter 
from which these elements are retranslocated by the tree prior to abcission 
(Marschner, 1995). Similarly, the young, leaf-rich biomass of frequently 
pruned trees has higher nutrient concentrations than the more woody 
biomass of infrequently pruned trees, although the quantity of biomass 
produced decreases with pruning frequency (Duguma et al., 1988). Roots 
cut off during soil tillage would also be expected to have higher contents 
of certain nutrients than roots that die naturally, although the question of 
nutrient retranslocation from senescing roots requires further clarification 
(see Section 12.4). Trees will also have higher nutrient concentrations in 
their biomass when they grow in nutrient-rich than in nutrient-poor soil 
(Budelman, 1989; Palm, 1995). Compilations of nutrient concentrations 
in tree biomass can be found in Young (1997), Palm (1995), books on 
animal nutrition and in the Organic Resource Database developed by the 
Tropical Soil Biology and Fertility Programme (TSBF) and Wye College, 
London ( 

The amount of nutrients contained in tree prunings has often been 
compared with the nutrient requirements of the crops to which the 
prunings were applied. Such comparisons are useful to illustrate the 
relative importance of the nutrient fluxes through the tree biomass, but it 
must be kept in mind that in simultaneous agroforestry systems a large 
part of the added nutrients might have also been available to the crops in 
the absence of the trees. According to Palm (1995), the nutrients in 4 Mg 
ha 1 of leaves from any of four leguminous tree species could meet the 
requirements of a maize crop for nitrogen, calcium and much of the 
magnesium and potassium. Nitrogen was not supplied in sufficient 
quantity to meet crop demands by two non-leguminous trees, and 
phosphorus was not supplied in sufficient quantity by any of the tree 
species studied. This latter point is also valid for other annual crops 

134 C.Schroth 

(Schroth et al., 1995c). Biomass is usually a less efficient source of 
phosphorus than of other macronutrients. 

Synchrony and synlocation of nutrient release with plant uptake 

When biomass decomposes on or in the soil, the nutrients may either 
remain in the soil in mineral form, be incorporated in the soil biomass and 
soil organic matter (immobilization), be taken up by plants, or be lost from 
the system through leaching or in gaseous form. The relative importance 
of these different pathways depends on the respective nutrient, the 
decomposing material, and the biotic and abiotic conditions under which 
the decomposition process takes place. It has been hypothesized that the 
efficiency of the uptake of nutrients released from biomass by crops can 
be increased by improving the synchrony and synlocation (i.e. the 
temporal and spatial correspondence) of nutrient release with plant 
uptake. This concept can be applied to nutrient release both from biomass 
and from soil organic matter and is particularly relevant where losses of 
released nutrients are likely to be high, such as in humid tropical regions 
with high risk of leaching and denitrification during most of the year, or 
in savanna regions with a pronounced flush of nitrogen mineralization at 
the beginning of the rainy season (Myers et al., 1994; Heal et al, 1997). 

Nutrient release from biomass and soil organic matter, and thus its 
synchrony and synlocation with crop uptake, can be influenced by 
management. Nutrient release from biomass can be controlled by selecting 
plant materials with desirable nutrient release kinetics and adjusting the 
timing and method of their application, with nutrient release being faster 
from soil-incorporated than from surface-applied materials. Nutrient 
release from soil organic matter is affected by the method and timing of 
soil tillage, mulching to reduce wetting and drying cycles, and the use of 
cover crops. To be effective in improving the uptake of a limiting nutrient 
by crops, it is important to consider whether synchrony and synlocation 
of release either directly from tree biomass or from soil organic matter is 
most relevant, so that appropriate management measures can be designed. 
This underlines the importance of studies of the fate of nutrients following 
their release from biomass. 

When litter or mulch is exposed to rainfall on the soil surface, a soluble 
nutrient fraction is rapidly leached from the biomass into the soil either 
in mineral or organic form. This fraction includes most of the potassium, 
and for freshly cut (as opposed to naturally fallen) materials it may also 
include a relevant percentage of the phosphorus in the biomass (Babbar 
and Ewel, 1989; Schroth et al. , 1992). Rapid initial release of nitrogen from 
Cajanus cajan prunings has been attributed to leaching (Schroth et al., 
1992), but this has not been confirmed for other materials (Vanlauwe et 

Decomposition and Nutrient Supply from Biomass 135 

al., 1995). The fate of such rapidly released nutrients, and especially 
whether they are taken up by crops or washed further downward in the 
soil, should depend on the presence and activity of root systems in the soil 
at this time and on pedoclimatic conditions. Up to now such studies tracing 
the fate of released nutrients have concentrated largely on nitrogen, and 
little information is available on what happens to other nutrients including, 
for example, the relatively large quantities of potassium leached from tree 
litter or primings. 

The release of nitrogen from biomass and its fate in the soil-crop 
system has received considerable attention, both because of its importance 
to plant growth in many agroecosystems and the availability of suitable 
tracers which facilitate such research. Data compiled by Palm (1995) show 
that the recovery of biomass nitrogen by the crops to which it is applied is 
generally less than 20% and often closer to 10%. This compares with values 
of 50-70% for mineral fertilizers, although fertilizer use efficiency in Africa 
is often much lower, and 20-30% for manure (Finck, 1992; Giller and 
Cadisch, 1995). Studies with 15 N-labelled legume biomass have shown that 
most of the nitrogen not recovered in the crops is incorporated into soil 
organic matter, from which it is then successively released through 
mineralization (Ladd, 1981). Accordingly, additions of legume tree 
prunings during 8 years of hedgerow intercropping caused substantial 
increases in soil nitrogen mineralization rates in Costa Rica (Haggar et al, 
1993), and nitrogen mineralization rates in the soil were also substantially 
increased under a leguminous cover crop in an Amazonian agroforestry 
system (Schroth et al. , 200 lc). Such results suggest that the nitrogen benefit 
to crops from biomass comes mainly from the longer-term build-up of 
readily mineralizable soil organic matter pools rather than from the short- 
term release of nitrogen from the decomposing material (Myers et al., 
1994; Palm, 1995). Consequently, it may be more important to improve 
synchrony and synlocation of nitrogen uptake by plants with its 
mineralization from soil organic matter than with its release from biomass, 
and this should be reflected by research priorities (see also Section 5.2). 
Little is known about the relative importance of short-term release as 
opposed to incorporation into mineralizable soil organic matter pools for 
phosphorus and sulphur (see Sections 5.3 and 5.4). 

Prediction of nutrient release from biomass 

As noted before, the dynamics of nutrient release from decomposing 
biomass under agroforestry conditions and the factors influencing it are 
the object of intensive studies. The principal objective of this research is 
to identify parameters that allow quantitative prediction of the time course 
of nutrient release from biomass as influenced by attributes of the biomass 

136 C.Schroth 

itself (the resource quality) and the physicochemical environment in which 
it is decomposing. As a rule, the decomposition of organic materials in soil 
leads to an initial net mineralization of nitrogen if the C:N ratio is <20, 
and to an initial net immobilization of nitrogen through incorporation of 
nitrogen from the surrounding soil into the decomposer biomass if the 
C:N ratio is >30. During the decomposition process, carbon is released as 
carbon dioxide, and the C:N ratio is progressively reduced below the 
threshold value for net mineralization. For phosphorus and sulphur, the 
corresponding threshold values are C:P<200 and C:S<200 for initial net 
mineralization, and C:P>300 and C:S>400 for initial net immobilization. 
These values apply if carbon and the respective nutrient are in compounds 
with similar degradation rates (Stevenson and Cole, 1999). 

For the prediction of nitrogen release from litter and primings of 
agroforestry tree species, the C:N ratio alone has not been found 
satisfactory, as several other chemical characteristics of biomass affect the 
decomposition dynamics, especially their lignin and polyphenol content 
(Mafongoya et al., 1998). Lignin, which is a low-quality substrate for 
decomposers, physically protects cellulose and other cell wall constituents, 
thereby reducing their degradation. A threshold level of 150 g kg- 1 lignin 
has been suggested above which decomposition is impaired. Polyphenols 
include hydrolysable and condensed tannins. Insoluble condensed tannins 
bind to cell walls and proteins and make them physically or chemically less 
accessible to decomposers. Soluble condensed and hydrolysable tannins 
react with proteins and reduce their microbial degradation and thus 
nitrogen release (Mafongoya et al., 1998). A critical soluble polyphenol 
content of 40 g kg 1 above which nitrogen release patterns are affected has 
been identified (Palm et al., 2001). The relationship between polyphenol 
contents and nitrogen dynamics can be improved by measuring the 
protein -binding capacity of the polyphenols (Handayanto et al., 1994). 
Both lignin monomer analogues and tannins have also been shown to 
directly inhibit the synthesis and activity of cellulolytic enzymes 
(Sinsabaugh et al. , 1991). 

Several indices of mass loss and nitrogen release from biomass have 
been proposed. Of these the (lignin + polyphenol):N ratio has been 
suggested as the most robust for materials commonly used in agroforestry, 
although its application still suffers from methodological difficulties in the 
polyphenol measurements. Such indices are not constant for plant species, 
because nitrogen, lignin and polyphenol contents vary for the same species 
with plant part, age and growth conditions including soil fertility and water 
supply (Mafongoya et al., 1998). Furthermore, there is evidence of rapid 
increases in tannin contents in the foliage of some trees following damage 
(Leng, 1997). For leguminous nodules, the ammonium and hexosamine- 
N contents were the best predictors of nitrogen release during 
decomposition (Wardle and Greenfield, 1991). Like the elemental contents 

Decomposition and Nutrient Supply from Biomass 


Characteristics of Organic Resource 


N > 25 g kg- 


Lignin < 150 g kg -1 
Phenol < 40 g kg -1 

Lignin < 150 g kg -1 


directly with 
annual crops 


Mix with 
fertilizer of 
high quality 

Mix with 
fertilizer or 
add to compost 

Surface apply 
for erosion and 
water control 

Fig. 6.1. 'Researcher's decision tree' for optimum use of biomass as a function of 
its quality (reproduced with permission from Palm et al., 1997). 

of biomass, the concentrations of organic constituents change during 
decomposition; for example, polyphenol contents of different litter types 
have been shown to decrease relatively rapidly (Pereira et al., 1998). A 
decision tree for the optimum use of biomass as a function of nitrogen, 
lignin and polyphenol contents has been proposed, where high-quality 
materials are directly incorporated in the soil as a nutrient source for 
annual crops, while materials of intermediate quality are applied together 
with fertilizer or high-quality biomass or are composted, and slowly 
decomposing, nutrient-poor materials are applied to the soil surface for 
erosion control (Fig. 6.1). In the absence of chemical analyses, the quality 
characteristics can also be estimated from simple field tests (Fig. 6.2). 

Leaves fibrous (do not crush) 
Highly astringent taste 

Leaves crush to powder when dry 









directly with 
annual crops 

Mix with 
fertilizer of 
high quality 

Mix with 
fertilizer or 
add to compost 

Surface apply 
for erosion and 
water control 

Fig. 6.2. 'Farmer's decision tree' for optimum use of biomass as a function of its 
quality (reproduced with permission from Ciller, 2000). 

138 C.Schroth 

A further potentially relevant measure of biomass quality is the soluble 
carbon content, which may be important for initial effects on microbial 
growth, and thus nutrient mineralization and immobilization. Its amount 
in plant materials is usually < 15% (Palm and Rowland, 1997), but may be 
as high as 34% in foliage of Eucalyptus spp. (Bernhard-Reversat, 1993). In 
the latter work, mineralization rates were higher for soluble carbon from 
Eucalyptus spp. than from Acacia mangium foliage, indicating that effects of 
this carbon fraction on microbes may differ between species, depending 
on its chemical nature. Relationships between soluble and insoluble 
fractions of various litter types and carbon dioxide release during the initial 
stages of decomposition have been studied by Bernhard-Reversat (1998). 
The soluble carbon fraction in plant debris may also contain substances 
with phytotoxic and fungistatic properties (Ramamoorthy and Paliwal, 
1993; Blum et al., 1999). 

Management of decomposition and nutrient release 

There are several management options for manipulating decomposition 
and nutrient release patterns (Mafongoya et al., 1998). These include the 
timing of biomass application and the incorporation of biomass into the 
soil, which accelerates decomposition, while surface application provides 
protection of the soil surface for longer. Of interest is also the possibility 
of mixing different litter types, such as recalcitrant materials, which 
provide a long-lasting surface mulch, with high-quality materials, which 
decompose and release nutrients rapidly. This can best be achieved by 
planting the respective plant species in association. Mixtures of litter types 
often have mineralization patterns equal to the weighted average of the 
patterns of the two separate materials (Palm et al., 2001), but non-additive 
effects of biomass mixtures on decomposition and nutrient release rates 
have also been observed in many studies (Chapman et al, 1988; Blair et 
al, 1990; Briones and Ineson, 1996; Wardleda/., 1997; Sakalarfa/., 2000). 
The mixture of high-quality Gliricidia sepium primings with prunings of 
Peltophorum dasyrrachis, which contains large concentrations of soluble 
polyphenols, led to complexation of proteins and unexpectedly low 
nitrogen release and nitrogen recovery by maize (Handayanto et al., 1997). 
Mixing nitrogen-poor maize residues with senesced leaves of pigeonpea 
(Cajanas cajan) also resulted in more prolonged nitrogen immobilization 
than would have been expected from the individual components (Sakala 
et al., 2000). Biomass may also be applied together with mineral nutrient 
sources to overcome nutrient imbalances (e.g. low phosphorus levels in 
biomass) and avoid initial immobilization of nutrients such as nitrogen and 
phosphorus when biomass with low concentrations of these elements (e.g. 
cereal straw) is applied to the soil (see Figs 6.1 and 6.2). The synergistic 

Decomposition and Nutrient Supply from Biomass 139 

effects of biomass and mineral fertilizers as nutrient sources have been 
reviewed by Palm et al. (1997). 

Litter effects on soil acidity and nutrient mobility 

The effect of different litter types on pH and nutrient mobility in acid soils 
has been the topic of recent studies (Noble and Randall, 1999). These 
authors found a highly significant relationship between the ash alkalinity 
of the plant materials and the pH shift of an acid soil during an 8-week 
incubation with ground leaves. Ash alkalinity was determined either by 
ashing the plant material followed by titration of the ash, or as the 
difference between the total cations and anions in the biomass. It was 
closely related to the calcium concentration in the samples, which could 
be used to estimate alkalinity. Based on ash alkalinity and the ability of the 
litter products to form stable organometallic complexes, which increase 
the mobility of nutrient cations in the soil, the authors suggested an index 
for the potential impact of plant materials on the base status of soils. 
Materials with: 

• high alkalinity and high complexing ability will have a high risk of 
nutrient losses (but also potential for subsoil improvement); 

• high alkalinity and low complexing ability are likely to increase the 
base status of the surface soil (the most favourable combination); 

• low alkalinity and low complexing ability are unlikely to affect the soil; 

• low alkalinity and high complexing ability will have a strong 
podzolizing effect (the least favourable combination). 

Litter effects on soil organisms 

Besides the nutrient release from biomass, the microbial and faunal 
decomposer community has also been the object of studies which have the 
potential to contribute to improved management practices in agroforestry. 
The composition and activity of this community depends on characteristics 
of the biomass and environmental factors. The microbial decomposer 
community differs between surface-applied and incorporated crop 
residues, and this may affect carbon retention in soil organic matter 
(Holland and Coleman, 1987). The faunal communities in soil and litter 
of agroforestry systems are also influenced by the quantity and quality of 
litter and their small-scale distribution (see Chapter 16). Faunal 
communities not only may influence the path of decomposition and 
nutrient release (Tian et al., 1995b), but may also affect soil properties such 

140 C.Schroth 

as soil structure and soil organic matter dynamics. Future research in this 
area should focus on how biomass can be used strategically to improve soil 
conditions through the stimulation of beneficial faunal and microbial 
groups (see Chapter 16). 

The animal pathway 

Instead of being directly applied to the soil as a source of nutrients and 
organic carbon, tree prunings are also often of interest as protein-rich 
fodder for domestic livestock. The resulting farmyard manure may be 
used for fertilizing crops, thereby recycling part of the nutrients in the 
biomass. Many tropical crops respond to manure applications with 
increased yields (Webster and Wilson, 1980). Whether biomass is better 
applied to the soil or used as fodder depends on the materials in question 
and on the degree of integration of crop and livestock production in the 
farming system. Within the season of application, nitrogen-rich biomass 
from trees such as Calliandra calothyrsus will release more nitrogen when 
applied directly to the soil than would be released from the manure 
resulting from feeding it to ruminants. In contrast, processing materials 
with low nitrogen but high decomposable carbon contents, such as 
barley straw, through an animal may lead to lower nitrogen immobilization 
when the faeces are applied to the soil than would be the case with fresh 
plant material (Delve et al., 2001). In a study in western Kenya it was 
concluded that feeding Calliandra calothyrsus biomass to diary cattle and 
using the manure for soil improvement was economically more 
advantageous than applying the biomass directly to a maize crop (Jama et 
al, 1997). 

The factors that determine the transformation of biomass into soil 
organic matter have been discussed in Chapter 4. 

6.2 Methods for Biomass and Nutrient Input with Litter 

Soils under agroforestry receive biomass inputs in the form of natural 
litterfall and primings. The quantification of nutrients in prunings is 
straightforward. The material should be subdivided into more or less 
homogeneous groups with respect to nutrient concentrations and moisture 
content, such as leaves, small wood (usually <2 cm), larger branches, and 
flowers and fruits. These groups should be weighed, subsampled and 
analysed separately. Methods for quantification of biomass and nutrient 
inputs in litter have been described in detail by Anderson and Ingram 

Decomposition and Nutrient Supply from Biomass 141 

Litter traps 

Tree and shrub litter is usually collected in litter traps, which may be 
constructed with plastic netting (mosquito netting is often used), mounted 
on a wooden or metal frame. A common collector size is between 0.25 and 
1 m 2 (Anderson and Ingram, 1993). The collectors are placed at a few 
decimetres above the ground to avoid contamination of the litter by soil 
or animals. This also accelerates drying of the collected litter, thereby 
slowing down decomposition processes. The traps must allow drainage of 
rainwater, but the openings must be small enough to retain small leaflets 
and litter fragments (< 1 mm). The leaflets of certain legume tree species 
are too small to be quantitatively retained by these commonly used 
collectors. In such cases, either tissues with smaller openings or a second 
layer of netting placed at an angle of 45° to the first can be used 
(F. Bernhard-Reversat, personal communication). Losses can be estimated, 
and eventually correction factors determined, by mounting the collector 
tissue on buckets or other closed recipients (Schroth et at., 1995b). 

Where the vegetation is more or less homogeneous and spatial 
patterns of litterfall are not of interest, the collectors can be placed 
randomly in the measurement plot. For obtaining a 5% standard error 
about the mean, 20 collectors or more per plot may be necessary 
(Anderson and Ingram, 1993). To improve the representativeness of the 
plot mean obtained (or to reduce the number of replicate samplers without 
losing precision), the positions can be changed from time to time as 
recommended for throughfall collectors (Lloyd and Marques, 1988). In 
studies where spatial variability of litter and nutrient inputs is of interest, 
permanent positions may be more useful. 

Agroforestry plots are usually not homogeneous and so systematic or 
stratified random sampling designs are usually preferable to random 
designs in litterfall studies (see Section 3.4). A systematic design for a 
planted fallow with trees at 2 m x 2 m would be to place collectors of 
1 m x 1 m size in a way that the sampling area covers one quarter of the 
space between four neighbouring trees with one corner of the collector 
attached to a tree (see discussion of smallest representative units in Section 
3.4). For other planting designs, the collector form, size and location 
should be chosen so that the sampled area covers a representative section 
of the plot. Other agroforestry plots are best subdivided into more or less 
homogeneous subplots, in which collectors are placed in a random pattern 
(stratified random sampling). For isolated trees (or trees of a certain species 
within a multispecies canopy), samplers should be used that cover a sector 
of a circle with the tree in the centre and the outer limit of the collector 
reaching 1-2 m beyond the limits of the tree crown. Preferential wind 
directions should be considered when deciding about the sampling design. 
For large trees, the angle of the sector that is covered by the collector can 

142 C.Schroth 

be reduced at some distance from the trunk to limit the total size of the 
collector. Small trees often produce only a small amount of litter, so to 
obtain reliable data it is often necessary to collect all of it (F. Bernhard- 
Reversat, personal communication). 

Litter traps measure the litter input into the soil, which is not always 
the same as the litter production by the tree crowns covering this soil. Litter 
typically falls during windy weather, and the leaves of tall trees can travel 
considerable distances before they reach the ground (or the collector). In 
continuous vegetation, such as forests, this may not cause errors, but it can 
lead to underestimation of litter production in small, isolated experimental 
plots or from isolated trees in savannas, as some of the litter may fall too 
far from the tree to be collected. Also, litterfall measurements in one plot 
may be influenced by the litter of neighbouring plots. 

Depending on the research question, the collected litter may have to 
be sorted according to species. In general, it should be separated into 
leaves, small woody litter (twigs <2 mm, bark), reproductive structures, 
and trash (sieve fraction <5 mm) (Anderson and Ingram, 1993). 
Subsamples should be dried at 60-80°C for the determination of dry 
weight and subsequent nutrient analyses. Wood samples may require 
drying at higher temperatures (105°C). 

If nutrient analyses are intended, the litter should be collected from 
the traps at least every 2 weeks to reduce leaching and decomposition, 
although more frequent collections are recommended for litter types that 
decompose rapidly and under high-rainfall conditions. Under dry 
conditions, less frequent collections may be sufficient, although 
contamination by dust or animals may occur (Anderson and Ingram, 
1993). When comparing nutrient concentration data from different litter 
studies, the collection interval and resulting experimental errors may have 
to be taken into account (Cuevas and Medina, 1986). 

Other methods 

Quantification of coarse litter such as branches or palm fronds requires 
larger sampling areas. As contamination with soil and decomposition losses 
are less of a problem than with leaf litter, the sampling area can be marked 
on the ground without recourse to a collecting device. The concept of 
smallest representative unit of a plot is useful for this (see Section 3.4). 
Litter fractions that are measured in the small collectors should be 
removed from the coarse litter before weighing. The litter is then also 
subsampled for the determination of dry weight (Anderson and Ingram, 

The litter production of herbaceous vegetation such as grassland is 
more difficult to measure than that of trees, because collectors cannot be 

Decomposition and Nutrient Supply from Biomass 143 

installed. A suitable method for measuring litter production and 
disappearance is the paired plot technique (Moore and Chapman, 1986). 
Biomass and nutrient inputs with litterfall and primings in various 
production systems in the humid tropics have been reviewed by Szott et 
al. (1991). 

6.3 Methods for Decomposition and Nutrient Release from 

The processes of litter decomposition and associated nutrient release, 
either under agroforestry conditions or in natural tropical ecosystems, 
have attracted the attention of many researchers, and numerous 
discussions of methods and results are available in the literature. Recent 
work has concentrated particularly on chemical characteristics of the litter 
or biomass that influence decomposition and nutrient release, and on the 
effect of decomposer organisms and environmental factors on these 
processes. The principal difficulty in decomposition experiments is that 
the material has to be identified during the decomposition process with 
minimal disturbance to effects of environmental factors, such as 
microclimate or access of decomposers, on decomposition (Anderson and 
Ingram, 1993). 


Decomposition of leaves and other materials of small size is usually studied 
in mesh bags with varying mesh sizes. Litterbags are made from plastic 
materials, but, where termites feed on the plastic or are to be excluded 
from the litter for experimental reasons, the mesh can be stainless steel or 
brass. The mesh size should be about 5-10 mm, which is small enough to 
minimize loss of fragments but large enough to allow access by 
invertebrates and to reduce alteration of the microclimate. Litterbags with 
finer mesh on their lower side and coarser mesh on their upper side are 
a good compromise, especially when small leguminous leaflets are under 
study (Lehmann et al., 1995). Care should be taken that the litter is 
arranged in a similar way in the bag as in the natural litter layer to imitate 
its microclimatic conditions. Litterbags can be buried in the soil when the 
decomposition of ploughed-in materials is under study, or can be 
suspended at some distance above the soil for monitoring the 
decomposition of leaves of large prunings in a mulch layer that remain 
attached to the branch and decompose without having contact with the 
soil (Gupta and Singh, 1981). 

144 C.Schroth 


Litterbaskets are an alternative to litterbags, and are supposed to create 
less artificial conditions than litterbags, while retaining more experimental 
control than working with unconfined litter (Blair et ai, 1991; Anderson 
and Ingram, 1993). They have a diameter and height of 1-3 dm and are 
made of metal or plastic mesh. The baskets are partly buried in the soil 
and can be filled with either disturbed soil, for arable conditions, or 
undisturbed soil, for uncultivated soils. The organic material whose 
decomposition is to be studied may be placed in the baskets either in layers 
for mulch or litter, or mixed with the soil for ploughed-in green manure, 
crop residues or roots. Processing the samples after the incubation is 
obviously more labour-intensive than with litterbags placed on the soil 
surface. The method also allows study of the effects of the decomposing 
litter on the soil (Lehmann et ai, 1998b). 

Unconfined litter 

For decomposition studies of larger woody materials, these can be marked 
or attached to a tether for easier localization and reweighed after certain 
time periods. For wood, the loss in density provides a means for 
monitoring the decomposition process with little disturbance to the sample 
and has been linearly correlated with carbon dioxide evolution (Yoneda 
et ai, 1977). In a study in the central Amazon, wood decomposition as 
measured from weight loss and respiration rate decreased with increasing 
wood density and increased with its moisture content (Chambers et ai, 

Where litter is released in pulses and can be clearly distinguished on 
the soil from older litter, or where no older litter is present, decomposition 
can be monitored by periodically quantifying the litter lying on the soil. 
This method can be useful for studying decomposition of crop residues 
after the harvest, or in systems where mulch is applied at regular intervals 
as in hedgerow intercropping or in systems with regularly pruned shade 
trees. It should, however, be noted that weight loss of unconfined leaves 
is a measure of litter breakdown rather than of its mineralization (Woods 
and Raison, 1982). If it is only the effect of added biomass on the soil, but 
not the transformation of the biomass itself that is to be monitored, then 
the biomass can be mixed with soil, from which samples are periodically 
taken for analysis (e.g. of mineral nitrogen). Ground leaf material has been 
used in this type of experiment to study chemical leaf characteristics 
independently from their physical properties (Palm and Sanchez, 1991). 

Decomposition and Nutrient Supply from Biomass 145 

Root decomposition 

In studies of root decomposition, the compromise between the avoidance 
of artificial conditions and the need to identify the decomposing material 
is particularly problematic. Unfortunately, accurate decomposition rates 
are of central importance for certain models for estimating root turnover 
(see Section 12.4). The most frequently used method for measuring 
root decomposition is the litterbag technique, although it seems to 
underestimate root decomposition rates in comparison with other 
methods. Beside the difficulty of collecting sufficient root material in the 
correct state of senescence and diameter class, the removal of rhizosphere 
constituents when preparing the roots for the incubation is a potential 
source of error (Vogt et a!,., 1991). In any case, the roots in the litterbags 
should be mixed with soil to ensure a good root-soil contact. 

As an alternative to litterbags, the decomposition of senesced roots of 
annual crops can be monitored in situ by sequential soil coring after the 
crop harvest, if weed growth is controlled or if dead weed roots can be 
distinguished from those of the crops (Schroth and Zech, 1995b; Lehmann 
and Zech, 1998). For perennial crops and trees, an equivalent situation is 
created with the trench plot technique. Soil monoliths, from which the 
vegetation is removed, are isolated from the surrounding soil, for example 
with plastic sheets, thereby cutting off roots in the monolith and impeding 
the ingrowth of new roots. Monoliths of variable sizes have been used, for 
example 1 m deep around a 1 m x 3 m plot (McClaugherty et al., 1984), 
and 30 cm deep around a 20 cm x 20 cm plot (Santantonio and Grace, 
1987). The decomposition of the roots in the monolith is monitored by 
sequential soil coring. The decomposition of tree roots in trench plots is 
generally about twice as fast as in litterbags. Problems of the trench plot 
technique include higher soil moisture compared with the surrounding 
soil (possibly increasing decomposition), and retarded decomposition of 
larger severed roots with significant carbohydrate reserves. There is also 
evidence that the presence of living roots may influence decomposition. 
It is not possible to decide which of the two methods produces more 
reliable results from present knowledge (Vogt et al., 1991), and so the use 
of a combination of methods is advisable. 

Technically more demanding than the aforementioned techniques are 
methods that depend on carbon isotopes for measuring the transfer of 
carbon from roots to soil, and thus root decomposition. This can be done 
after labelling roots with 14 C, either in the laboratory (Jones and Darrah, 
1994) or in the field (Swinnen et al., 1995), and measuring the 14 C0 2 
evolution from the roots or temporal changes of the distribution of the 
isotope between roots and soil (Cheshire and Mundie, 1990; Swinnen et 
al, 1995). Alternatively, the input of root carbon from C 4 species in soil 
previously dominated by C 3 species (or the other way round) can be 

146 C.Schroth 

quantified by comparing treatments where either root and shoot biomass 
or only root biomass is returned to the soil (Balesdent and Balabane, 1996). 
These techniques have not yet been used with trees, although they seem 
to be appropriate in principle. Minirhizotron observations have also been 
used to determine in situ root decomposition rates (see Section 12.4). 

Litter selection and pretreatment 

The litter for decomposition studies has to be chosen in agreement with 
the objectives of the experiment. Where the decomposition of primings is 
to be assessed, fresh leaves should be used. Where the decomposition of 
naturally fallen litter is to be studied, freshly fallen leaves should be 
collected, which have lower contents of certain nutrients (nitrogen, 
phosphorus, potassium) and soluble organic carbon than green leaves and, 
therefore, decompose more slowly (Woods and Raison, 1982). Similar 
considerations apply to roots: for decomposition studies of roots that have 
been cut off during tillage, living roots can be collected, but, for natural 
root turnover, senescent roots would be required, which decompose at 
very different rates from severed living roots (Publicover and Vogt, 1993). 
Senesced roots are, however, difficult to collect in sufficient quantity. 
Drying of litter before the incubation should be avoided as this may 
introduce artefacts (Anderson and Ingram, 1993), unless the materials are 
also subjected to drying and rewetting under natural conditions for 
decomposition, which is often the case in the tropics. 

Nutrient release from decomposing litter 

Nutrient release from decomposing litter is usually measured by collecting 
litter samples from litterbags or tethered litter at intervals and then 
measuring their weight and nutrient concentration and expressing it 
relative to the initial values. Alternatively, nutrient release from 
undisturbed litter can be monitored by collecting the leachate beneath the 
litter with trays or funnels and subtracting the nutrient input in rainfall 
or throughfall (Woods and Raison, 1982). This approach allows separation 
of nutrients released in organic and inorganic forms, but does not take 
into account gaseous nutrient losses from decomposing biomass (Janzen 
and McGinn, 1991). The two approaches can also be used in combination. 
Tubes filled with mixtures of soil and litter which are periodically leached 
and the leachates analysed provide a means to study nutrient release and 
immobilization dynamics under controlled conditions with a high temporal 
resolution (Sakala et al., 2000). 

Nutrient release studies from decomposing roots are complicated by 

Decomposition and Nutrient Supply from Biomass 147 

the problem of soil contamination, which is greater for fine than for coarse 
roots and tends to increase as decomposition advances. According to Misra 
(1994), soil contamination can account for up to 28% of the potassium in 
a root sample. The amount of soil adhering to the roots can be estimated 
from ashing (Misra, 1994) or from the determination of total carbon in the 
sample, assuming that a soil-free root has a carbon content of about 45% 
(Schroth and Zech, 1995b). The nutrient concentration of the soil can be 
determined by digesting a separate soil sample with the same method used 
for the root samples. Misra (1994) also found that washing of roots resulted 
in potassium losses of 24% and that the nutrient losses further increased 
when the root samples were stored in water after separation from the soil. 
Similar rules presumably apply to other types of biomass that become 
contaminated with soil during decomposition studies. Minimum contact 
with water and immediate drying of samples after their separation from 
soil are generally recommended. 

Particularly useful for agroforestry research are decomposition studies 
with 15 N-labelled biomass as these allow tracking of the incorporation of 
released nitrogen into vegetation (trees and crops) and different 
components of the soil organic matter (HaggaretaL, 1993; Xuetal., 1993; 
Vanlauwe et al, 1998a). For problems of pool substitution with 15 N-labelled 
biomass which can sometimes lead to large underestimates in nitrogen 
recovery see Section 7.3 and McDonagh et al. (1993). 

Measuring the role of decomposer groups and environmental factors in 

The contribution of different decomposer groups, such as microorganisms 
and invertebrates of different sizes, to litter decomposition can be assessed 
by using litterbags of different mesh sizes. For exclusion of all invertebrates, 
a mesh of 20-60 [im is required (Woods and Raison, 1982). The efficiency 
of exclusion should, however, be checked as invasion of 20 [im litterbags 
by microarthropods, presumably through oviposition or juvenile forms, 
has been reported (Vreeken-Buijs and Brussaard, 1996). Takamura and 
Kirton (1999) measured wood decomposition with an open tray design 
that they claimed excluded most termites while permitting access of most 
other invertebrates. Decomposer groups can also be excluded by adding 
biocides (such as naphthalene for invertebrates) to the incubated material 
(Heneghan et al., 1999), but these may have side effects on non-targeted 
organisms (Blair et al, 1991). Alternatively, certain decomposer groups 
can be selectively added to incubated litter to study their effect on litter 
breakdown (Tian et al., 1995b). Woods and Raison (1982) proposed to 
separate total weight loss of unconfined litter into weight loss due to 
reduced leaf area (presumably caused by fragmentation or consumption 

148 C.Schroth 

by larger invertebrates) and weight loss per unit leaf area (presumably 
caused by decomposition by microbes and microfauna). The use of 
enzymic techniques for the mechanistic study of litter decomposition 
processes has been reviewed by Sinsabaugh et at. (1991). 

Monitoring environmental factors such as rainfall and moisture 
content of the litter during decomposition experiments may help to 
explain observed patterns of decomposition and nutrient release 
(Gupta and Singh, 1981; Reddy, 1992). Factors that potentially limit 
decomposition can be studied experimentally by adding them to 
decomposing litter in the field and measuring the effect on decomposition 
or decomposer organisms in comparison with an untreated control, as was 
done with water by Cornejo et at. (1994) and certain nutrients by Cuevas 
and Medina (1988). 

Decomposition constants 

Various functions can be fitted to decomposition and nutrient release data. 
The most common one is the single exponential model W t = W e~'", where 
W and W t are the mass or quantity of nutrient at the beginning of the 
experiment and after time t, respectively, and k is the decomposition or 
nutrient release constant. The time after which the material loses half its 
initial mass is obtained as t m = In2/A . As biomass usually consists of fractions 
that decompose and release their nutrients rapidly, and other more 
recalcitrant ones, double or higher exponential models have also been 
used that have coefficients for the proportion of material in each fraction 
and their respective decay constants. Linear models may well describe the 
initial phase of decomposition. These and other models and their 
respective advantages are discussed by Wieder and Lang (1982). 

Decomposition standards 

For the comparative assessment of decomposition processes at different 
sites, the inclusion of standard materials in decomposition experiments is 
useful. Proposed materials include birch (Betula spp.) sticks (Anderson and 
Ingram, 1993) and Leucaena leucocephala biomass (Budelman, 1988). 
Artificial substrates have also been used (Woods and Raison, 1982). 

6.4 Measures of Resource Quality 

The term resource quality is now prefered to the synonymous term 
substrate quality. It is a summary expression for the intrinsic characteristics 

Decomposition and Nutrient Supply from Biomass 149 

of an organic material that, together with the decomposer community and 
environmental factors, determine the speed of its decomposition and 
nutrient release (Swift et al., 1979; Heal et a!,., 1997). In a wider sense, 
characteristics that influence the fate of the carbon and nutrients derived 
from decomposing biomass, such as their incorporation into soil organic 
matter, may also be included in the term resource quality (Palm and 
Rowland, 1997). Both chemical and physical properties influence the 
quality of a resource, although the chemical ones seem to be generally 
more important and have been studied more thoroughly. Chemical and 
physical substrate characteristics can be separated experimentally by 
studying the decomposition of either the whole or ground materials 
(Agbim, 1987; Palm and Sanchez, 1991). 

Methods for characterizing chemical resource quality have been 
reviewed by Palm and Rowland (1997), and the following recommenda- 
tions are mostly taken from that paper. The chemical quality of an organic 
material is influenced both by its carbon constituents (carbon quality) and 
by its nutrient content and the chemical form of the nutrients (nutrient 
quality). Palm and Rowland (1997) proposed a minimum set of resource 
quality characteristics for decomposition studies, including lignin, soluble 
carbon, total nitrogen, total phosphorus, ash-free dry weight, and soluble 
phenolics if total nitrogen is greater than 1.8%. Measurement of the 
protein-binding capacity to assess reactive polyphenols may also be useful 
in some cases. For some characteristics, the choice of the analytical method 
is critical as different methods may produce different results, which leads 
to problems of comparability between studies. As for any laboratory 
analysis, the inclusion of reference materials that are exchanged with other 
laboratories is recommended. 

Biomass samples are collected in more or less homogeneous units (e.g. 
old leaves, young leaves, branches), and the contribution of each unit to 
the total biomass is quantified. For the analysis of soluble compounds, the 
samples are dried at 35-40°C, otherwise a drying temperature of 60-80°C 
can be used (105°C for woody materials). All measurements should be 
referred to dry weight as obtained at this temperature. The dried samples 
are ground to pass a 1 mm sieve. The recommended procedure and 
methods for biomass analyses are given in Fig. 6.3. The alternative 
procedure in Fig. 6.4 still requires testing before routine application (Palm 
and Rowland, 1997). 

Physical biomass characteristics that could influence decomposition 
include toughness, particle size and surface properties such as waxiness. 
In a litter layer, sclerophyllous leaves often have less contact with the soil 
than soft leaves, and this may influence their water content and 
decomposer activity. However, chemical and physical litter characteristics 
are not independent, and useful physical measures of litter quality have 
not yet been identified (Palm and Rowland, 1997). 


C. Schroth 

Plant material 
(dried 35-40"C, 1 mm mesh) 

Boiling water 
(TAPPI, 1988) 


50% aq. methanol 
2-5 mg mh 1 (Anderson 
and Ingram, 1993) 


Weight loss 

(Waterman and Mole, 1 994) 

Simple sugar Kje | dah | 

(Dubois era/., 1956) 


Acid-detergent CTAB 
(van Soest, 1963) 

Acid-detergent fibre 

72% H,SO, 

ADF residue 


soluble N 



Residue ash 



Fig. 6.3. Recommended procedures for analysis of carbon proximate fractions 
using three separate extractions (reproduced with permission from Palm and 
Rowland, 1997). CTAB, cetylmethyl ammonium bromide; ADF, acid-detergent fibre. 

Plant material 
(dried 35-40'C, 1 mm mesh) 

50% aq. methanol 

2-5 mg ml -1 

(Anderson and Ingram, 1993) 


Simple sugar 
(Dubois etal., 1956) 



soluble N 


Weight loss 

(Waterman and 
Mole, 1994) 



Extract acid- 
detergent CTAB 
(van Soest, 1 963) 

Acid-detergent fibre 

72% H 2 SQ 4 


ADF residue 


Residue ash 


Fig. 6.4. Alternative procedure for analysis of carbon proximate fractions using two 
separate extractions (reproduced with permission from Palm and Rowland, 1997). 

Chapter 7 
Nutrient Leaching 

J. Lehmann 1 and G. Schroth 2 

1 College of Agriculture and Life Sciences, Department of Crop and 
Soil Sciences, Cornell University, 909 Bradfield Hall, Ithaca, NY 
14853, USA; 2 Biological Dynamics of Forest Fragments Project, 
National Institute for Research in the Amazon (INPA), CP 478, 
6901 1-970 Manaus, AM, Brazil 

7.1 Synopsis 

Nutrient leaching is the downward movement of dissolved nutrients in the 
soil profile with percolating water. Nutrients that are leached below the 
rooting zone of the vegetation are at least temporarily lost from the system, 
although they may be recycled if roots grow deeper. Leached nutrients 
may contribute to groundwater contamination in regions with intensive 
agriculture. Nitrate leaching is also a significant source of soil acidification. 
In humid climates, some nutrient leaching occurs even under natural 
vegetation, but agricultural activities can greatly increase leaching losses 
(Havlinrfa/., 1999). 

Soil and climatic factors that influence nutrient leaching 

In general, water transport below the rooting zone requires that the soil 
water content exceeds field capacity and the water balance is positive, 
which means that water inputs with rainfall (and irrigation) exceed 
evapotranspiration. Therefore, nutrient losses through leaching are 
generally higher in humid than in dry climates (Havlin et ah, 1999). In 
certain soils, however, water can infiltrate into the subsoil through 
continuous vertical macropores when the bulk soil is dry. This is especially 
important in cracking clay soils (Vertisols) at the onset of the rainy season 
(Smaling and Bouma, 1992). Macropores are also created by faunal activity 
and root growth. They only conduct water under conditions of heavy 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 151 

F.L. Sinclair) 

152 J. Lehmann and C. Schroth 

rainfall or irrigation, under other conditions they are filled with air. 
Macropore or bypass flow may increase nutrient leaching following the 
surface application of fertilizers, because a solution with high nutrient 
concentration then infiltrates rapidly into the soil with little contact with 
the soil matrix. On the other hand, macropore flow may also protect 
nutrients present in smaller soil pores from being leached by rapidly 
channelling away surplus water (Cameron and Haynes, 1986; van 
Noordwijk^a/., 1991b). 

Soils with high water infiltration rates and low nutrient retention 
capacity, such as sandy soils and well-structured ferrallitic soils with low- 
activity clays and low organic matter contents, are particularly conducive 
to nutrient leaching (von Uexkiill, 1986). Some nutrients are easily leached 
from organic soils (see below). Subsoil acidity also tends to increase nutrient 
leaching by restricting the rooting depth of sensitive plants (see Section 

In the subsoil of many tropical soils the mobility of nitrate and other 
anions decreases because of increasingly positive net charge and, therefore, 
anion retention by soil minerals, and this increases the probability that 
these ions are eventually taken up by deep-rooting plants (see Box 8. 1 on 
p. 17 1). It is, therefore, important to distinguish between nutrient leaching 
within the soil profile, from the topsoil into the subsoil, leading to 
temporary nutrient loss, and leaching beyond the rooting zone of deep- 
rooting plants, into the groundwater, leading to permanent nutrient loss. 

Susceptibility of different nutrients to leaching 

The leaching risk for a nutrient increases with its mobility in the soil. 
Among nutrient anions, nitrate is particularly easily leached because it 
shows negligible interaction with the negatively charged matrix of most 
topsoils and is, therefore, very mobile in the soil (see Section 5.2). 
Nitrification rates are variable in tropical soils, but can be sufficiently high 
to make nitrate the dominating form of mineral nitrogen even in acid soils 
(Robertson, 1989; Schroth et al., 1999a). As a consequence, leaching may 
contribute significantly to negative nitrogen balances of agricultural 
systems (Smaling et al, 1993). In seasonal climates, nitrate is also 
particularly exposed to leaching because a mineralization flush of organic 
nitrogen that causes release of large quantities of nitrate in the topsoil often 
occurs when dry soil is rewetted at the onset of the rainy season, at a time 
when crops have not yet been sown or are still small (Birch, 1960). 

A mineralization flush at rewetting of dry soil has also been reported 
for sulphur (Havlin et al., 1999). Sulphate is also readily leached from 
surface soils, the losses being highest in soils dominated by monovalent 
cations (potassium, sodium) and lowest in soils with high amounts of 

Nutrient Leaching 153 

aluminium (Havlin et al., 1999) (see Section 5.4). Dissolved organic sulphur 
contributed between 18 and 86% of total dissolved sulphur at 2 m depth 
in an agroforestry system in central Amazonia (J. Lehmann, unpublished 

In contrast to nitrate and sulphate, phosphate is immobile in most soils 
because of precipitation and adsorption to mineral surfaces, and leaching 
is therefore negligible, except in certain very sandy and organic soils (Wild, 
1988) (see Section 5.3). Dissolved organic phosphorus forms are more 
mobile in soil than phosphate (Havlin et al., 1999). Phosphorus may also 
be lost if surface soil particles are eroded in runoff (see Section 17.1). 

The percolating soil solution that carries nutrients down the soil profile 
is necessarily electrically neutral; therefore, anions are leached together 
with equivalent amounts of cations. In most soils, the cations most likely 
to be leached are calcium and magnesium. In West African savanna soils, 
close relationships between the combined concentration of Ca and Mg and 
that of nitrate in the soil solution below the crop rooting zone have been 
reported, pointing to the role of fluxes of nitrate, and to a lesser extent 
chloride, as factors controlling calcium and magnesium leaching in these 
soils (Pieri, 1989). In sandy soils, considerable amounts of magnesium can 
be leached after applications of potassium chloride or potassium sulphate 
fertilizers (Havlin et at., 1999). Potassium is usually leached in much smaller 
quantities than calcium and magnesium even when applied as fertilizer 
and was not related to nitrate fluxes in the aforementioned studies in West 
Africa (Pieri, 1989). However, significant potassium leaching may occur in 
sandy and organic soils and in high-rainfall areas (Malavolta, 1985; Havlin 
et at., 1999). Among the micronutrients, manganese and boron are 
susceptible to leaching in certain soils (Havlin et at., 1999). 

Management practices that reduce nutrient leaching 

A number of agricultural practices reduce nutrient losses through leaching 
by increasing the synchrony and synlocation of nutrient uptake by the 
vegetation with nutrient supply from soil, mineral fertilizers and organic 
materials (see Section 6.1). These include: 

• early sowing of crops at the onset of the rainy season in savanna 
climates to make use of the mineralization flush of nitrogen upon 
rewetting of the soil (Myers et at., 1994); 

• rapid installation of a vegetation cover after forest or fallow clearing 
to avoid nutrient losses from bare soil (Webster and Wilson, 1980; von 
Uexkull, 1986); 

• applying fertilizers (especially nitrogen) in several small applications 
during the cropping season rather than all at once; and 

154 J. Lehmann and C. Schroth 

• placing fertilizer at the zone of maximum root activity of tree crops 
(IAEA, 1975; Havlin etal., 1999). 

Leaching of nutrients from organic sources 

Of particular relevance for agroforestry is the efficient management of 
nutrients in organic materials, including biomass and manure, for 
increased crop uptake and reduced leaching losses (see also Chapter 6). 
Nutrient release from organic sources is generally more difficult to predict 
than from mineral fertilizers and so developing practices to counteract 
leaching is particularly important. Nutrients are often released from 
organic sources at a time when there is little crop uptake and consequently 
more opportunity for leaching. Although leaching losses of nutrients from 
organic sources comparable to or even higher than from mineral sources 
have been reported (Havlin et al., 1999), other results show lower leaching 
of nutrients from biomass than from mineral fertilizer. Snoeck (1995) 
applied 15 N-enriched urea or biomass from either Leucaena leucocephala or 
Desmodium intortum that was also enriched with 15 N to coffee plants on an 
Oxisol in Burundi and measured the distribution of the nitrogen in 
undecomposed biomass, coffee plants and soil after 1 year. Almost half of 
the urea nitrogen was lost from the system, presumably by leaching below 
30 cm soil depth, but most of the nitrogen released from biomass was 
retained in the topsoil (Fig. 7.1). Lehmann et al. (1999c) found that 
sorghum (Sorghum bicolor) took up more nitrogen from labelled 
ammonium sulphate than from Acacia saligna leaves in a runoff 
agroforestry system in northern Kenya. Much of the fertilizer nitrogen 
that was not taken up by the crop was lost from the system by leaching or 
volatilization, whereas 99% of the biomass nitrogen was recovered in soil 
and crop at the end of the cropping season. This highlights the important 
point that labile nutrients are both more vulnerable to leaching and more 
readily taken up by crops, so in some circumstances farmers may tolerate 
higher leaching losses from mineral fertilizers because the short-term 
nutrient uptake by crops and crop yields may also be greater than from 
organic nutrient sources. Mechanisms responsible for lower leaching losses 
from biomass than from mineral sources include: 

• slower nutrient release, which is especially important when relatively 
large quantities of nutrients are applied at a time, as in the latter study, 

• stimulation of microbial growth in the soil by organic nutrient sources, 
leading to temporary immobilization of nutrients in the microbial 

Nutrient Leaching 155 

Fig. 7.1. Distribution of 15 N-labelled nitrogen 1 year after application to coffee 
(Coffea arabica) plants as urea or biomass of Leucaena leucocephala or 
Desmodium intortum on an Oxisol in Burundi. Total quantities applied were 9.2 g 
per plant of urea-N vs. 34.4 g per plant of Leucaena-N (1225 g biomass with 
2.81% N) and 20.5 g per plant of Desmodium-N (682 g biomass with 3.01% N). 
Note that absolute quantities of nitrogen taken up from urea and Leucaena 
biomass were similar (2.1 g per plant); differences in percentage uptake resulted 
from different quantities applied. Nitrogen uptake from Desmodium biomass was 
1 .2 g per plant (after Snoeck, 1 995). 

Effect of trees on nutrient leaching 

One of the central hypotheses of agroforestry is that the continuous or 
intermittent presence of trees in land-use systems can increase the 
efficiency with which nutrients are retained in the soil-plant system and 
transformed into biomass and harvested products instead of being lost by 
leaching (Young, 1997). This hypothesis has been confirmed in a limited 
number of studies. Seyfried and Rao (1991) measured lower nutrient 
concentrations in the soil solution and calculated lower nutrient leaching 
in a multistrata agroforestry system with cocoa, banana and Cordia alliodora 
than in a maize monocrop in Costa Rica. Horst (1995) reported lower 
nitrate concentrations in the soil solution and consequently less leaching 
under hedgerow intercropping with Leucaena leucocephala and annual food 
crops than in the agricultural control treatments in southern Benin. 

156 J. Lehmann and C. Schroth 

Lehmann et al. (1999a) measured lower nutrient leaching under an Acacia 
sa/igwa-sorghum intercrop than under pure sorghum with runoff 
irrigation in northern Kenya. 

Several mechanisms may contribute to reduced nutrient leaching 
under agroforestry compared with agricultural monocrops. Through 
increased litter, mulch and root production, agroforestry practices may 
contribute to increased soil organic matter levels and therefore increased 
cation exchange capacity and nutrient retention (see Chapter 4). Also, trees 
may create macropores with their roots or through the stimulation of 
macrofaunal activity (see Chapter 16), and this may help to channel surplus 
water through the soil with limited contact with nutrients in the soil matrix 
(bypass flow, see above and Chapters 10 and 11). These tree effects are 
desirable in both fallow rotations and simultaneous agroforestry systems. 
Furthermore, water uptake by trees may reduce water infiltration and, 
therefore, nutrient leaching. Lower soil water contents in agroforestry 
plots than in agricultural controls have often been reported (Malik and 
Sharma, 1990; Rao et al., 1998). Trees may also reduce nutrient 
concentrations in the percolating soil solution through nutrient uptake 
(Horst, 1995). Reduction of nutrient leaching by trees through uptake of 
water or nutrients is only desirable in fallow systems and in tree-crop 
associations during periods when no crops are present, such as before 
sowing and after the harvest of annual crops. When both trees and crops 
are present in a field at the same time, water and nutrient uptake by trees 
may reduce nutrient leaching, but may also cause yield depressions of the 
crops through competition. These conflicting effects of trees in 
simultaneous agroforestry systems are apparently one reason why 
agroforestry associations such as hedgerow intercropping have often been 
successful in maintaining soil fertility at higher levels than agricultural 
controls, but have not improved crop yield (Rao et al., 1998) (see also 
Chapter 5). 

The safety-net hypothesis proposes that it is possible to achieve 
reduced nutrient leaching without increased root competition between 
trees and crops (van Noordwijk et al, 1996). The hypothetical safety net 
for leached nutrients is formed by trees that possess few superficial roots, 
but whose deep roots spread laterally below the rooting zone of associated, 
shallow-rooting crops. Here, in the subsoil, they intercept nutrients and 
water, thereby reducing nutrient leaching without being associated with 
too much competition with the crops in the topsoil. The safety-net concept 
is an idealization that guides the search for tree species that exhibit a high 
degree of niche differentiation with crops in that they have relatively 
uncompetitive root systems in surface soil, but are sufficiently deep-rooting 
to acquire relevant amounts of subsoil resources. Utilization of subsoil 
nitrogen by a deep-rooting, uncompetitive tree species, Peltophorum 
dasyrrhachis , in association with groundnut on an Ultisol in Sumatra has 

Nutrient Leaching 157 

been demonstrated, where the tree obtained more than 40% of its nitrogen 
from below the crop rooting zone (Rowe et al, 1999). Further support for 
the hypothesis comes from the observation of stratified root systems of 
associated plant species in different natural and artificial ecosystems, which 
develop either because the vertical root distribution of associated species 
responds differently to soil and climatic factors, or because the root system 
of one species avoids competition from the other species through increased 
growth in the subsoil (Schroth, 1999). However, a clear experimental 
demonstration of a safety-net effect in terms of lower nutrient leaching in 
agroforestry than in agricultural systems that is not offset by competition 
for water and nutrients in the top soil is still lacking. 

Another way in which agroforestry (or intercropping) techniques can 
reduce nutrient leaching is by optimizing the spatial patterns of nutrient 
use in tree crop plantations. If tree crops are planted at final spacing, they 
often take several years to fully occupy the soil with their root systems, and 
during this time nutrients released from the soil, a cover crop or 
decomposing residues from the previous vegetation may be leached in the 
spaces between the tree crop plants unless these are occupied by suitable 
intercrops, shade trees or spontaneous vegetation. Nitrate leaching in the 
spaces between 5-year-old widely spaced tree crops (peach palm - Bactris 
gasipaes, cupuagu - Theobroma grandiflorum, Brazil nut - Bertholletia excelsa 
and annatto - Bixa orellana) has been reported from an Amazonian Oxisol, 
indicating the presence of surplus nutrients and water that could be used 
for additional crop production (Schroth et al, 1999a). On a similar soil, 
the nitrate distribution in the soil under a 15-year-old oil palm plantation 
indicated that interplanting with shade-tolerant crops may still have been 
viable at this age (see Fig. 8.1 on p. 170). Similarly, in mature plantations 
of coconut (Cocos nucifera) significant quantities of light, water and nutrients 
may not be captured by the tree crops and so can be utilized by intercrops 
without reducing tree crop yield (Mialet-Serra et al., 2001). The potential 
of tree crop-based agroforestry systems to reduce nutrient leaching has 
been discussed in more detail by Schroth et al. (2001b). 

In contrast to simultaneous tree-crop associations, fallow systems rely 
on the rapid development of deep roots of the fallow trees to intercept 
nutrients in the subsoil which were leached during the previous cropping 
phase (see Chapter 8). Modelling results suggest that, under high-leaching 
conditions and for very mobile nutrients such as nitrate, fallows may not 
be effective in recycling leached nutrients; instead, the permanent 
presence of tree roots as in simultaneous systems would be necessary to 
intercept these nutrients before they are leached too deep into the subsoil 
(van Noordwijk, 1989). As discussed in Box 8.1 onp. 171, anion retention 
in the subsoil of many tropical soils increases the potential for intermittent 
fallows with deep-rooting trees to recycle leached nitrate, suggesting that 
leaching losses could be reduced at a lower competition cost than with 


Lehmann and C. Schroth 

permanent tree-crop associations. Experimental comparisons of 
simultaneous and sequential agroforestry systems under different 
pedoclimatic conditions with respect to nutrient cycling and overall 
productivity would be required to confirm this. 

7.2 Methods for Soil Solution Composition 

The soil solution comprises the soil water and the inorganic and organic 
substances that it contains. Nutrient leaching can be determined from the 
quantity and composition of the soil solution that percolates to depths 
greater than the rooting depth of the plant species present. Depending on 
the respective research question, the collection of soil solution may be 
necessary either below the rooting zone of an annual crop, which could be 
anything from a few decimetres to a few metres deep, or below the rooting 
zone of a tree, which may be many metres deep. For technical reasons, 
solution sampling below the deepest roots of agroforestry trees is often not 
feasible. However, measurements of the soil solution composition and 
nutrient leaching at different depths under crops and trees may still 
provide valuable information on the ability of the plant species to capture 
nutrients from the percolating solution in the subsoil. 

There are three approaches to the measurement of nutrient leaching 
through the collection and analysis of soil solution (Fig. 7.2). 

• Solution samples are collected and leaching is estimated from the 
concentration of dissolved nutrients and the movement of the solution 
in the soil, which is measured separately. Suction cups are the most 
common tools for soil solution sampling. A more recent development, 
the tensionic sampler, allows soil solution measurements without 
application of suction due to diffusion of solutes through a porous cup 


Suction cups 

Lysi meters 

Resin cores 

Fig. 7.2. Schematic depiction of the installation of suction cups, lysimeters and 
resin cores. The arrows indicate the direction of solution movement through the 
soil and into the collectors. 

Nutrient Leaching 159 

into the sampling device. As no suction is applied, the same device can 
be used for tensiometer readings (Moutonnet et al., 1993). Alternative 
methods include the extraction of solution from field-moist soil by 
centrifugation or displacement with different immiscible liquids, but 
these techniques require destructive soil sampling for every solution 
measurement and are thus not suitable for determining cumulative 
nutrient losses during longer time intervals. Examples from 
agroforestry experimentation using suction cups and a water balance 
for leaching measurements include Seyfried and Rao (1991), Kiihne 
(1993) and Lehmann et al. (1999a). 

• The percolating solution is collected in lysimeters with a defined 
collection area, so that the downward flux of the solution can be 
determined from the collected sample volume, which is also used for 
nutrient analysis. A principal difference exists between tension 
lysimeters (including suction plates) and free-draining lysimeters. An 
agroforestry application of the latter technique is described by Santana 
and Cabala-Rosand (1982). 

• The percolating solution passes through a sampler with a defined area, 
in which nutrients are retained on an ion-exchange resin; in this 
method the percolating water volume does not need to be measured. 
This inexpensive and simple technique is still under development. 
Agroforestry applications include Hagedorn et al. (1997) and 
Lehmann rf a/. (1999b). 

Suction cups 

Suction cups are fine-porous filters that are closed at one end and mounted 
on a rigid or flexible tube at the other end. The dimensions of the cups 
vary in general between 3 and 60 cm in length and between 0.5 and 10 
cm in diameter. The soil solution is sucked through the cups and into a 
collection bottle by applying a vacuum, either temporarily or continuously. 
The vacuum applied to suction cups has to be adjusted in order to sample 
only the most mobile water, although it needs to be strong enough to 
gather enough soil solution for analysis. This can be a problem in very 
clayey soils. Battery-powered pumps can continuously keep the vacuum 
at the desired level. Where the soil water content shows pronounced 
fluctuations during the sampling, periodic adjustments of the applied 
vacuum maybe necessary. Self-regulating devices are now available, which 
measure the soil water suction and automatically adjust the applied 

Suction cups are well suited for studies of short-term fluctuations and 
small-scale variability of soil solution chemistry, because the solution can 
be collected at short time intervals, and even the installation of a large 

160 J. Lehmann and C. Schroth 

number of cups at depths of several metres causes little disturbance in a 
plot. Shortcomings of suction cups are that the mobile soil solution is not 
adequately sampled in most cases and that the collected solution cannot 
be related to a defined soil volume or infiltration area (Grossmann and 
Udluft, 1991). Preferential water flow through macropores may constitute 
the majority of the percolating solution, but its infiltration is often too rapid 
to be sampled with suction cups. Failure to sample the initial mobile soil 
water after rewetting may result in large errors in estimates of leaching, 
as this water may be enriched with nutrients (Lord and Shepherd, 1993). 
The suction cup technique may yield reliable data in sandy, unstructured 
soils (Webster et al., 1993), whereas clayey and well-structured soils may 
pose considerable difficulties. 

Various materials are used for the cups, which differ in their purpose 
and applicability. Cups made from P80 ceramic material with a pore 
diameter of about 1 |lm are frequently used because of their low price. In 
these cups, however, significant amounts of phosphate and dissolved 
organic matter may be retained in the material. If organic compounds are 
of interest, therefore, the suction cups have to be conditioned for a long 
time in the soil being studied. More inert materials for phosphorus 
sampling include Teflon, glass, polytetrafluoroethene (PTFE)/quartz or 
cellulose acetate fibres (Dorrance et a!,., 1991; Beier and Hansen, 1992). 
Pretreatment of the cups with dilute hydrochloric acid, followed by 
distilled water, before installation is generally necessary (Angle et al., 1991; 
Beier and Hansen, 1992). This also eliminates problems with aluminium 
desorption from P80 cups. The glues connecting the cups with shafts may 
leak organic substances but contamination problems can be avoided by 
reducing the contact zone of solution and glue or by using glue-free suction 
cups. The installation should be done well in advance of the intended 
measurements to allow for chemical equilibration of the cups with the 
surrounding soil and to let the soil settle after the installation (Lord and 
Shepherd, 1993; Webster et al., 1993). 

To avoid microbial transformations, the collected solution should be 
frequently removed from the collection bottles and should be either 
analysed immediately or deep-frozen (Angle et al., 1991). In practice, the 
solution often remains in the collection bottles in the field (dark bottles are 
preferred to prevent the growth of algae) or in the suction cups for a few 
hours or days before it is collected. Chemicals can be added to the bottles 
to inhibit microbial activity, such as chloroform or various acids. These 
additions, however, may interfere with the intended measurements of pH, 
certain colorimetric reactions, and dissolved organic matter. Alternatively, 
the solution can be collected without preservation and be digested prior 
to the analysis of elements that could be affected by microbial growth 
(especially nitrogen and phosphorus). In this case, only total fluxes of these 
nutrients are determined without specifying the chemical compounds. 

Nutrient Leaching 161 

Collection of the solution from several samplers in the same bottle or 
pooling of collected solution prior to analysis are common practices to 
obtain more representative samples without increasing the number of 
analyses. However, pooling may increase the risk of losing the whole 
sample if one cup produces contaminated solution. It is thus advisable to 
analyse samples from all individual cups at the beginning of a 
measurement programme. 

When installing the cups, care has to be taken to avoid preferential 
water flow along the shafts by sealing with clay or rubber discs. At shallow 
depths, the cups are often installed at an angle to the soil surface. At greater 
depths, they may be installed horizontally from a soil pit. After long-term 
use, the connectors and tubings of the suction cups may show signs of 
ageing and not hold the vacuum any more. Termites and ants may destroy 
plastic tubes, and polyethylene materials are also susceptible to light 
damage. In soils rich in dispersible clay, the cup pores can become clogged 
after prolonged use, in which case the cups have to be replaced. 


Lysimeters as defined here are horizontally installed trays that capture 
percolating soil water. Discussions of the technique can be found in 
Dorrance et al. (1991), Barbee and Brown (1986) and Russell and Ewel 
(1985). Free-draining lysimeters are open at the top and rely on gravity 
for collecting the percolating solution (Jordan, 1968). In contrast, tension 
lysimeters (or suction plates) consist of porous materials similar to those 
used in suction cups with a sealed bottom, to which a vacuum is applied. 
This vacuum should ideally resemble the matrix potential of the 
underlying soil so that the water percolation is not influenced by the 
sampler. Unlike suction cups, lysimeters can also sample macropore flow. 

Free-draining lysimeters have been found to yield more soil solution 
than suction cups in clayey soils (Barbee and Brown, 1986). However, they 
will still underestimate percolation because the soil water has to overcome 
the soil matrix potential to enter the collector, and in finely textured soil 
with weak soil structure no water at all may be collected. The sampling 
technique may also influence the solution chemistry. For example, free- 
draining lysimeters were found to yield solution from a wetting front later 
and with higher calcium and potassium concentrations than suction cups, 
which collected a solution with higher nitrogen and magnesium 
concentrations (Marques et al, 1996). In an unstructured soil, similar 
results for both methods have been obtained (Webster et al, 1993). 

In experiments with annual crops or before tree planting, lysimeters 
can be installed from the soil surface below the plough layer. For greater 
installation depths and with an established tree root system, the installation 

162 J. Lehmann and C. Schroth 

has to be carried out from the side using a soil pit. Special care has to be 
taken to ensure a good contact between the lysimeter and the overlying 
soil (Barbee and Brown, 1986). The same rules as for suction cups apply 
to selection and pretreatment of the porous material and preservation of 
the collected solution. 

Resin cores 

Resin cores are cylinders filled with ion-exchange resin and installed in 
the soil in a way that allows water percolation (Schnabel, 1983). Nutrients 
in the percolating solution are adsorbed to the resin and can be extracted 
after the core has been removed. Similarly to lysimeters, resin cores have 
a defined surface area, so that a time integral of nutrient leaching per unit 
area is obtained. Depending on the leaching rate and the construction of 
the cores, they can be left in the soil for periods as long as 6-12 months. 
Disadvantages of this technique are that collection of the resin is destructive 
and that short-term leaching events are not detected due to the integrative 
nature of the measurement. 

The cores usually have a length and diameter of 5-20 cm and can be 
built from PVC tubing. The water flux through the cores relative to the 
surrounding soil and hence the validity of the results will strongly depend 
on the similarity between resin cores and soil with respect to water 
permeability at different water contents. To increase the similarity of 
hydraulic properties between cores and soil, the resin should be mixed 
with acid-washed sand or soil (Hagedorn et al, 1997). Even with such 
precautions, it may be difficult to determine absolute leaching losses in 
finely textured soils. The collection efficiency of resin cores can be checked 
by tracer experiments using chloride. 

Commercially available exchange resins used for water purification 
can be used in the cores since they show low blank values and high 
recovery rates for nitrate, ammonium, phosphate and basic cations. 
However, the resin properties should be checked in preliminary 
adsorption experiments. Dissolved organic matter and organic nutrients 
are often difficult to determine with this method, because inexpensive 
resins may bleed organic compounds and the necessary quantities of 
analytical resins may not be affordable (Lehmann et al., 2001d). 

At the end of the measurement period, the resin cores are recovered 
and the resin is extracted with KC1 or CaCl 2 solution, depending on the 
nutrients to be analysed. The cores are cut and different resin layers 
extracted separately to make sure that the nutrient load of the percolated 
solution did not exceed the exchange capacity of the resin. The lowest 
layer may contain a large portion of nutrients derived from capillary rise 
and should be excluded from the calculation of nutrient leaching. 

Nutrient Leaching 163 

Resin cores should not be confused with resin bags, which are placed 
in the soil to obtain a time-integral of the availability of certain nutrients 
without an intention to measure nutrient leaching (Binkley, 1984). 

7.3 Tracer Methods for Nutrient Leaching 

Several different approaches exist for employing tracers to measure 
nutrient leaching. Their suitability depends on experimental objectives. 
First, it depends on the source of the nutrients whose leaching is being 
assessed. These maybe derived from precipitation, nutrient mineralization 
in the topsoil, mineral fertilizer or organic materials such as primings or 
manure. The source of the nutrient will determine the chemical form in 
which the tracer is best applied, for example, as a component of a mulch 
material or as a mineral fertilizer. Secondly, the choice of method will be 
influenced by the question as to which of the various processes that 
determine nutrient leaching need to be considered in the experiment. For 
example, different methods would be applicable if there were only interest 
in water percolation as influenced by the availability of water and the 
infiltration rate of the soil than if it were also necessary to understand 
nutrient uptake by plants. 

There are two different approaches to the measurement of nutrient 
leaching with tracers: 

• measuring the content of the tracer in the soil to a certain depth, 
comparing the recovered quantity with the amount that was applied 
(or measured in the same soil volume at an earlier date) and 
considering the difference as leached (for nitrate profiles see Box 8.1 
on p. 171); and 

• measuring the amount of tracer that passes through a certain soil 
depth (such as the maximum rooting depth of a crop) with the 
techniques described in Section 7.2. 

A large number of different tracers exist, which can be classified as 

• radioactive isotopes (e.g. 32 P, 33 P, 35 S); 

• stable isotopes (e.g. 15 N, 34 S); or 

• other tracers (e.g. strontium, lithium, rubidium, bromide or chloride). 

Radioisotopes and stable isotopes allow use of the same nutrient for 
which the leaching is to be determined without significantly changing its 
concentration in the soil. Non-isotopic tracers can simulate the behaviour 
of certain chemically related nutrients in soil, for example, chloride and 
bromide can be used for nitrate, strontium for calcium, and rubidium and 
lithium for potassium (see also Section 8.2). However, higher amounts of 

164 J. Lehmann and C. Schroth 

the tracers may have to be used to be detectable because of the low 
sensitivity of the analyses or sometimes the high amount of the substance 
already present in the environment. If nutrient leaching from organic 
sources or the effect of nutrient uptake by plants on nutrient leaching is 
to be studied, only radioactive or stable isotopes of nutrient elements 
can be used, because only these possess similar properties during 
decomposition and microbial transformation to the nutrient of interest. 
However, radioisotopes will in many cases be considered too dangerous 
for a field assessment of nutrient leaching, leaving stable isotopes as the 
only option. 

Tracers can be applied to the soil in mineral form or in organic 
materials such as mulch, litter, compost or manure. In the latter case, the 
substrate has to be labelled beforehand. This can be done by supplying 
the tracer to a plant whose litter or biomass is to be used in the experiment 
and collecting the respective materials after some weeks. The application 
method depends on the objectives. Labelled fertilizer will usually be 
applied in the same manner as unlabelled fertilizer. If the emphasis of the 
study is on the assessment of the nutrient transport in the soil rather than 
the release from a certain nutrient source, the tracer is best applied in 
solution, which can be applied more evenly than solid materials such as 
fertilizer. A simple hand-sprayer or electric pump can be used to spray a 
known amount of the dissolved tracer on a defined area. The amount to 
be applied will depend on the amount already present in the soil and the 
duration of the experiment. For example, 15 N can be applied at a dose of 
1 g m" 2 to be detectable even after several months in cropping systems with 

If the purpose of the study is to examine nutrient leaching after 
fertilization, large amounts of nutrients are usually applied to the soil. This 
may cause pool substitution of the tracer with native soil nutrients, as 
reviewed for nitrogen isotopes by Jenkinson et al. (1985). As a result, more 
unlabelled soil nitrogen may be leached with fertilization using a labelled 
nitrogen source than without, and total nitrogen leaching as affected by 
fertilization may, therefore, be underestimated. Only by measuring total 
nutrient losses together with tracer losses can the extent of the added 
nitrogen effect be determined and conclusions drawn about the leaching 
of applied and native soil nitrogen. 

Non-isotopic tracers are useful as long as nutrient uptake by plants 
does not need to be considered. For example, nutrient leaching at the 
onset of the rains before crop planting has been studied with bromide as 
a tracer in an East African Vertisol (Sticksel et al. , 1996). Walker et al. (1991) 
used the chloride content of wet deposition for measuring groundwater 
recharge as affected by land-use change in semiarid Australia. Also, 
hydrogen isotopes can be used to monitor water movements in soil 
(Miinnich, 1983). However, in the presence of vegetation, these tracers 

Nutrient Leaching 165 

would only reflect the effect of plant water uptake on leaching, but not the 
effect of actual nutrient uptake by plants from the percolating soil solution. 
Such data can be combined with measurements of nutrient concentrations 
in the soil solution (see Section 7.2) to determine nutrient leaching. 

The tracers can be retrieved for the analysis by collecting the soil 
solution (see Section 7.2), extracting the soil with a salt solution or a dilute 
acid (see section for the respective element in Chapter 5) or, in the case of 
nitrogen, be analysed directly by dry combustion of the soil (Barrie et al., 

Nitrogen occurs in the soil solution mainly as nitrate, ammonium and 
dissolved organic nitrogen (see Section 5.2). These nitrogen forms differ 
in their behaviour with respect to: 

• the soil matrix in terms of their adsorption and diffusion; 

• soil microorganisms in terms of nitrification, denitrification and 
mineralization; and 

• plant uptake. 

The analysis of 15 N is mostly done after the combustion of the entire 
soil sample. If separate analyses of the different nitrogen forms are of 
interest for the study, the sample has to be fractionated before the 15 N 
analysis. The separate analyses of nitrate- 15 N and ammonium- 15 N, for 
example, in soil solution samples or KG extracts, is possible after steam 
distillation (Bremner and Edwards, 1965; Buresh et al., 1982) or diffusion 
(Brooks et al, 1989). The distillates are freeze-dried or dried by diffusion. 
The dried samples can be directly analysed by isotope mass spectrometry. 
Distillation is more laborious and expensive than diffusion and subject to 
a higher analytical error since cross-contamination can easily occur. Several 
ways of reducing this problem have been discussed by Mulvaney (1986) 
and Mulvaney et al. (1994). 

In certain ecosystems, a large proportion of the total nitrogen in 
solution can be in an organic form, and it may then be interesting to 
consider organic nitrogen forms in leaching studies. The amount of 15 N 
in organic form can be obtained by subtracting the inorganic 15 N 
(ammonium + nitrate 15 N) from the total 15 N in the soil solution. Further 
fractionation into hydrophilic and hydrophobic organic nitrogen is also 
possible (Quails and Haines, 1991). 

7.4 Dyes as Tracers for Preferential Flow Paths 

Water infiltration and nutrient movements in soil are strongly influenced 
by macropores (see Section 7.1). The macroporosity of soil can be 
influenced by tree roots and burrowing soil animals, whose activity may 
be favoured by the presence of tree litter (see Section 10.1). Laboratory 

166 J. Lehmann and C. Schroth 

and field procedures for measuring soil porosity and water infiltration are 
discussed in Sections 10.4 and 11.4. Dyes are useful for staining flow 
pathways in the soil, so that these can be analysed in relation to other 
characteristics, such as the distribution of roots, faunal structures (van 
Noordwijk et a!,., 1993b) or soil microbial properties (Bundt et al., 2001). 
A disadvantage is that the infiltration of dyes is often limited to a few 
decimetres from the depth of application. Dyes that have been successfully 
used include rhodamine-B (Douglas, 1986), acid-red 1 (Ghodrati and Jury, 
1990), methylene blue (van Noordwijk et al., 1991b) and brilliant blue 
(Bundt et al, 2001). Van Ommen et al. (1988, 1989) detected preferential 
flow paths in soil to 70 cm depth by infiltrating a solution of iodine, which 
does not interact much with the matrix of most soils. The iodine is 
transformed into a coloured complex after treating the exposed soil with 
starch and Cl 2 . Dyes can be applied in the field by spraying on the soil 
surface, or through metal rings or bore holes as in infiltration 
measurements (see Section 11.4). Flow paths are evaluated in vertical 
trenches and/or by successively removing horizontal soil layers (Ghodrati 
and Jury, 1990). 

Chapter 8 
Nutrient Capture 


1 Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil; 2 College of Agriculture and Life Sciences, Department of 
Crop and Soil Sciences, Cornell University, 909 Bradfield Hall, 
Ithaca, NY 14853, USA 

8.1 Synopsis 

An ecologically important difference between trees and annual crops is 
that, due to their longer lifetime, trees can form larger root systems and 
can thus take up nutrients and water from a greater volume of soil. Where 
compact soil layers restrict the growth of annuals but less so of trees, this 
gives an additional advantage to the trees. Depending on species and 
environment, important differences between trees and annual crops may 
exist both in the vertical and in the horizontal extension of the root 
systems. Where tree roots reach deeper than the roots of annual crops, 
they may be able to take up nutrients and water from the subsoil which 
are not accessible to the crops, and this may improve tree growth without 
leading to a proportionate increase in competition with associated crops 
(see complementarity in Box 5.2 on p. 101). The same occurs in 
environments with patchy nutrient distribution in the horizontal plane, 
where far-reaching lateral roots may enable trees to tap nutrients at a 
distance which crops cannot acquire. This may include nutrient-rich 
patches in depressions that receive run-on water and fertile sediments 
along watercourses. 

There are different ways in which nutrients that have been taken up 
by trees from the deeper subsoil or fertile soil patches can become available 
to crops. When the trees are cut, for example at the end of a fallow, 
nutrients stored in their biomass may become available to crops provided 
that the tree material decomposes on site. (Where fallow vegetation is 
burnt, some of the nutrients will be recycled but a proportion, particularly 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 167 

F.L. Sinclair) 

168 G. Schroth and J. Lehmann 

of the nitrogen, will be lost; see Chapter 9.) Nutrients are also released 
into the soil from living trees through pruning, litterfall, dying roots or 
crown leaching and may thus become available to associated crops. 
However, the trees compensate for these nutrient losses through nutrient 
uptake from the soil. Therefore, crops are most likely to profit from 
increased nutrient capture by associated trees when they are efficient 
competitors for these nutrients once they have been released into the soil. 
Competitive crops could force associated trees to take up a substantial part 
of their nutrients from deep or laterally distant soil and could then 
scavenge a share of these nutrients when they are released into the soil 
from decomposing tree litter or primings (Schroth et al., 200 lb). Nutrients 
may also be recycled from trees when animals feed on tree fruits and 
foliage and then drop their excreta close to the trees. Where animals 
congregate under trees or birds perch on branches, they may concentrate 
nutrients close to the trees because of enhanced deposition of material 
derived from foraging in a wider area in the vicinity of the tree. This latter 
effect would not be called nutrient capture by trees, although the effects 
are similar. 

Nutrient capture from the subsoil: 'nutrient pumping' 

Nutrient capture by trees from the subsoil can include nutrients released 
by weathering of primary minerals and also nutrients leached from the 
top soil that are then recycled by the trees. Capture of newly weathered 
nutrients is restricted to relatively young soils where weatherable minerals 
still occur within the reach of tree root systems, including colluvial or 
alluvial soils with irregular nutrient distribution with soil depth. 

Despite the prominence of nutrient pumping by trees as a theoretical 
concept in agroforestry, the actual importance of the process in different 
ecosystems and agroecological situations is not very well documented. 
Studies on tree crops in seasonally dry tropical climates have shown that 
nutrient uptake from the subsoil increases when the surface soil dries out 
(Comerford et al, 1984), and the process could, therefore, be expected to 
be important in savanna areas. However, in West African savannas, Kessler 
and B reman (1991) assume that nutrient pumping by trees is of limited 
importance, because most trees present in the landscape are too shallow 
rooted (despite the occurrence of some very deep-rooted species), and 
many soils are either too shallow or too dry and nutrient-poor at depth to 
make nutrient pumping a useful strategy. Other authors, in contrast, have 
stressed the importance of deep-rooting trees with wide distributions in 
the region. Acacia Senegal, Acacia tortilis, Faidherbia albida and Azadirachta 
indica were all found to root down to the water table at from 16 m to 35 m 
depth in sandy soils in Senegal (Leakey et al., 1999) and there is evidence 

Nutrient Capture 169 

of large nitrate reserves at depth in groundwaters in arid zones (Edmunds 
and Gaye, 1997). Use of water from the deeper subsoil (>2 m depth) 
during the dry season has been demonstrated for some woody species in 
the Brazilian cerrado savanna (Jackson et a!,., 1999). In a savanna in Belize, 
on the other hand, pines (Pinus spp.) had tap roots, but other trees were 
shallow-rooted, and nutrient enrichment under their canopy could not be 
explained by nutrient recycling from the deeper subsoil (Kellman, 1979). 
Not surprisingly, where exotic tree species, selected for fast above-ground 
growth, have been grown with crops in seasonally dry environments, the 
trees have generally been shallow-rooted and competitive with associated 
crops (Sinclair, 1996). Clearly, nutrient capture at depth in natural and 
managed semiarid environments appears to be very site and species 
specific and it makes sense to look for complementarity in natural 
associations between tree and herbaceous species on particular site types 
to find candidate species for use in agroforestry. 

The recycling of leached nutrients from the subsoil by deep tree roots 
could be a relevant process in humid and subhumid regions, where nitrate 
derived from mineral fertilizer or organic sources is rapidly leached out 
of the topsoil together with nutrient cations such as calcium and 
magnesium during the rainy season (see Section 7.1). Soils with a high 
percentage of kaolinite and oxides of iron and aluminium, such as Oxisols 
and Ultisols, or soils of volcanic origin that are rich in allophane can have 
a significant anion exchange capacity that enables them to retain nitrate 
by sorption to the mineral phase. Anion exchange capacity (and nitrate 
sorption) increases with decreasing pH and decreasing organic matter 
content of the soil. They are, therefore, usually negligible in the topsoil 
and are most pronounced in acidic subsoil horizons. Nitrate sorption 
isotherms, describing the increase of nitrate sorption with increasing 
concentration of nitrate in solution, for different depths of Amazonian and 
Kenyan Oxisols can be found in Cahn et al. (1992) and Hartemink et al. 
(1996), respectively. 

By slowing the downward movement of nitrate, anion sorption 
increases the probability that part of this nitrate is taken up by deep roots 
and returned to the topsoil through litterfall, primings or leaching from 
the tree crowns. Nitrate accumulations have been reported from different 
tropical regions under annual crops (Cahn et al., 1993; Weier and Macrae, 
1993; Hartemink et al., 1996; Jama et al, 1998) and also under a 
coffee plantation in Kenya (Michori, 1993, cited in Buresh and Tian, 
1998), multistrata agroforestry and tree crop monocultures in central 
Amazonia (Schroth et al, 1999a, 2000a) and a runoff agroforestry system 
in semiarid northern Kenya (Lehmann et al., 1999a) (see also Box 8.1). 
Trees in humid tropical regions have often been seen as shallow-rooted, 
but it is now established that their roots can reach many metres deep 
(Nepstad et al, 1994; Canadell et al, 1996), provided that compact, acid 


G. Schroth and I. Lehmann 

or otherwise unfavourable subsoil conditions do not impede deep root 

Studies in western Kenya showed that planted fallows of fast-growing 
trees (Calliandra calothyrsus, Sesbania sesban) produced root length densities 
of > 0.1 cm cm- 3 to below 1.5 m soil depth and reduced soil nitrate at 0-2 
m depth by 150-200 kg ha -1 within 11 months after their establishment 
(Jama et al., 1998). They thereby recycled considerable amounts of subsoil 
nutrients, which could be made available to subsequent crops (or crops on 
neighbouring fields) through application of the tree biomass, either 
directly or after use as animal fodder. Factors that influenced these very 
promising results were certainly the fast growth of the trees, the high 
planting density of 1 m x 1 m, and the fact that the soils presented no 
obstacles for tree root development (Jama et al. , 1998). They thus illustrate 
the considerable potential of planted fallows, fodder banks and fuelwood 
plantations for nutrient recycling, but further research is necessary on the 
efficiency of trees in subsoil nutrient capture under other environmental 
and management conditions. Where tree spacing is wide, as in many 
tree-crop associations, tree root growth in the subsoil may be less than in 
closely spaced fallows because there is less intraspecific root competition 
in the topsoil. Also, the efficiency of widely spaced trees in nutrient 
recycling from the subsoil may be limited to a certain area close to the trees 
to which the deep tree roots have access (Fig. 8.1). Therefore, the 
evaluation of spatial patterns of root distribution and activity must be part 
of studies of nutrient leaching and recycling. Many tropical soils also 



f. 100 



& 150 H 


-•— 1 m tree distance 
-■— 2.5 m tree distance 
-*— 4 m tree distance 








Mineral N (mg kg soil) 

Fig. 8.1. Distribution of mineral nitrogen (nitrate-N + ammonium-N, means and 
s.e.) in the soil of an oil palm (Elaeis guineensis) plantation in central Amazonia at 
different tree distances, reflecting the restricted lateral root extension of the palms 
at depth (reproduced with permission from Schroth era/., 2000a). For 
corresponding root distribution data see Fig. 12.1. 

Nutrient Capture 


present impediments to tree root growth in the subsoil and hence nutrient 
pumping, such as hardened, very nutrient-poor or very acidic horizons 
(Kessler and Breman, 1991). Furthermore, there are indications that 
regular and frequent shoot pruning, which is a common management 
practice for trees in agroforestry, can reduce the rooting depth of trees, 
either through changes in root system architecture (van Noordwijk et a!,., 
1991a), or through a general reduction in root biomass. Such effects of 
shoot pruning have been found to be tree species specific (Jones et al., 
1998). All these factors need to be taken into consideration in future 
research on subsoil nutrient capture in agroforestry. 

Box 8.1. Nitrate profiles as indicators of nitrogen leaching 
and recycling. 

Leaching of nitrate and its recycling from the subsoil through deep-rooting plants 
are particularly relevant processes in agroforestry because nitrogen availability 
often limits crop yields, and nitrate leaching contributes to cation loss and soil 
acidification and may be a source of groundwater contamination (see Sections 5.2 
and 7.1). The easiest way of obtaining information about nitrate leaching and 
recycling is usually by collecting soil samples from different depths and extracting 
the nitrate as described in Section 5.2. Shepherd etal. (2000) compared the nitrate 

S- 2 


Good maize 
Poor maize 
Markhamla trees 
Banana plantation 
Weed fallow 
Eucalyptus woodlot 

12 3 4 5 6 

Nitrate-N (mg kg -1 soil) 

Fig. 8.2. The distribution of nitrate-N under seven land-use types and plant 
species on smallholder farms in Kenya illustrates the tighter nutrient cycles in 
systems with trees and perennial crops compared with annual maize 
(reproduced with permission from Shepherd etal., 2000). 


172 G. Schroth and I. Lehmann 

Box 8.1 . Continued. 

distribution to 4 m depth under different land-use types, including annual and 
perennial crops, trees and spontaneous fallow, in smallholder production systems 
in Kenya and found the largest accumulations of nitrate in the subsoil under maize 
(Fig. 8.2). Although such information does not allow quantification of nitrate losses 
from the different systems, the results demonstrate the advantage of integrating 
trees and perennial crops in annual cropping systems for closer nutrient cycles. 

Using a similar approach, Jama era/. (1 998) compared five tree species with 
respect to their ability to use nitrate that had accumulated in the subsoil during 
previous maize crops in Kenya as a criterion for their value in improved fallows. 
Nitrate profiles under different tree crops and a cover crop in a multistrata 
agroforestry system in Amazonia showed that the leguminous cover crop 
contributed more to nitrate leaching than three out of four tree crops in the system 
(Schroth ef a/., 1999a). Information on such small-scale patterns of nitrate 
distribution can help to optimize the design and management of complex cropping 
systems by showing where in a system nutrients are used inefficiently and where 
additional plants could be added or fertilizer rates reduced. Ideally, such 
information would be combined with quantitative nutrient balances for the whole 
system, as obtained, for example, in catchment studies. 

Nitrate profiles can also provide useful information about the soil volume 
from which crops or trees take up nutrients. The extension and shape of the rooting 
zone of trees, especially in the subsoil, is important information for evaluating their 
potential role in nutrient cycling in agroforestry systems. For example, for 
functioning as a safety net for nutrients leached from soil in which associated 
annual crops are growing (see Section 7.1), tree root systems need a sufficient 
lateral extension in the subsoil under the crop rooting zone (van Noordwijk etal., 
1996). Information on tree root distribution in the subsoil is difficult to collect 
through root studies and is, therefore, hardly ever available. However, by studying 
the spatial patterns of nitrate distribution in the soil, information about the lateral 
root extension at depth and the spatial patterns of nitrate leaching can be inferred 
with relatively little effort, as shown in Fig. 8.1 . A similar approach can be used 
to analyse the influence of scattered trees, boundary plantings and hedgerows on 
subsoil nutrients in agroforestry associations (Mekonnen etal., 1999). 

Repeated measurements of the nitrate distribution in the soil in the same area, 
for example at the beginning and the end of a cropping season or a fallow period, 
provide information on temporal changes of nitrate concentrations at different soil 
depths, and these may be related to nitrate leaching and uptake. Hartemink etal. 
(1 996, 2000) showed that nitrate in the subsoil (50-200 cm) of two Kenyan soils 
decreased under Sesbania sesban and weed fallows, but not under maize and 
concluded that subsoil nutrients and water were used more efficiently by the 
fallows than by the crops. 

Nitrate profiles do not usually allow measurement of nitrate fluxes in a strictly 
quantitative sense, because several processes which simultaneously influence 
nitrate concentrations at a given soil depth cannot easily be distinguished. Nitrate 
accumulations in the subsoil may be the result either of leaching of nitrate from 
the topsoil, or of in situ mineralization of nitrogen, or of a combination of both 
processes. Weier and MacRae (1 993) measured the nitrogen mineralization rates 

Nutrient Capture 173 

at different depths in an Australian soil where nitrate was known to accumulate 
in the subsoil below shallow-rooted crops. They concluded that most of the subsoil 
nitrate had been leached from the topsoil, although in situ mineralization of 
nitrogen could make a contribution to the nitrate accumulation in the subsoil under 
black gram (Vigna mungo). Similar information from other sites would be helpful 
in the interpretation of nitrate profiles. 

Decreases in subsoil nitrate content may also be caused by several processes, 
which may occur simultaneously. First, nitrate may be taken up by the vegetation, 
such as by deep-rooting trees. This requires that roots extend to the respective soil 
depth which should be checked in studies of nitrate capture (Jama et al., 1998). 
Secondly, during periods with downward water movement, nitrate may be leached 
to greater soil depths. Thirdly, nitrate may be lost through denitrification in the 
subsoil, a process that could itself be influenced by the presence of plant roots, 
since denitrification has been found to be limited by the availability of carbon 
rather than denitrifying microorganisms in certain subsoils (McCarty and Bremner, 
1993). However, studies of denitrification in tropical subsoils and the potential 
influence of deep tree roots are not available. 

Nitrate profiles can thus provide highly relevant information that can help to 
improve the nutrient cycles of agroforestry systems. However, because of the 
methodological problems outlined above they should usually be interpreted in 
largely qualitative terms. The analysis of nitrate profiles is greatly facilitated by 
complementary measurements of vertical and horizontal patterns of nitrogen 
mineralization, root distribution, soil water dynamics, and nitrogen uptake by the 

Horizontal redistribution of nutrients by trees 

In addition to the capture of subsoil nutrients, trees may also cause 
horizontal redistribution of nutrients by taking up nutrients from an area 
reaching far beyond the limits of their crowns through lateral roots. Part 
of these nutrients is deposited in the proximity of the trees through 
litterfall and crown leaching. Savanna trees may possess lateral roots 
extending several tens of metres from the trunk (Kessler and Breman, 
1991; Stone and Kalisz, 1991), and horizontal nutrient redistribution by 
these roots is assumed to contribute to the commonly observed nutrient 
enrichment in the soil under these trees (Kessler and Breman, 1991). 
Substantial lateral tree root extension has also been observed under 
agroforestry conditions: lateral roots of Senna siamea shelterbelts on shallow 
sandy soils in central Togo extended 20-30 m from the trunk, which was 
more than twice the tree height (Schroth et al., 1995a), observations on 
isolated Parkia biglobosa trees at three sites in Burkina Faso found 
substantial rooting up to 10 m from trees, which was almost 3 m beyond 
their mean crown radius (Tomlinson et al., 1998), and roots of Senna siamea 

174 G. Schroth and J. Lehmann 

and Dactyladenia barteri in a hedgerow intercropping experiment in the 
humid forest zone of Nigeria extended 15 m and 5 m from the trees, 
respectively (Hauser, 1993). Narrow alleys between hedgerows in 
hedgerow intercropping are usually entirely permeated by tree roots. This 
led to the observation that favourable tree effects on soil nutrient status 
may often have been overestimated in agroforestry experiments with 
relatively small plots adjacent to one another, because roots from trees in 
treatment plots may have taken up nutrients from neighbouring control 
plots that were supposed to be a sole crop (see Section 3.2). 

In many cases, horizontal redistribution of nutrients by trees cannot 
be seen as a positive effect of trees like vertical nutrient capture, because 
although it leads to higher tree growth and eventually higher yields of 
crops that profit from nutrient enrichment under the trees, this is done at 
the expense of neighbouring areas, which may include sheltered crop 
fields and pastures or even the neighbouring farm (van Noordwijk et al., 
1996). In other situations, however, trees may be able to capture nutrients 
from niches that would not be exploited by crops, such as stream banks, 
sites of old anthills where phosphorus has accumulated, or areas where 
livestock have been housed or home compounds, and to redistribute them 
into crop plots. 

Both the vertical and the horizontal capture of nutrients by trees are 
poorly quantified processes and, in view of their potential importance for 
nutrient cycling in agroforestry, quantitative studies on both of them are 
required. Of particular interest is how system design, species selection and 
management can be combined to increase vertical nutrient pumping and 
reduce lateral nutrient scavenging from crop fields by agroforestry trees 
(Schroth, 1995, 1999). 

8.2 Tracer Methods for Nutrient Uptake 

The soil volume from which roots are taking up nutrients can be 
determined in two ways: (i) by measuring the nutrient depletion in the 
soil, and (ii) by labelling certain areas in the soil with tracers and measuring 
the uptake into the plant. The first approach is a very crude measurement, 
because other processes like mineralization and leaching occur 
simultaneously with uptake, and nutrient depletion in a certain soil volume 
thus cannot be directly explained by nutrient uptake. Additionally, 
nutrient uptake by different plants in mixed cropping systems cannot be 
measured separately. 

The second approach provides the possibility of directly measuring 
the nutrient uptake by a plant from a certain soil volume such as at a 
certain depth in the subsoil or at a certain lateral distance from a tree, 
relative to the uptake from a reference soil volume such as the topsoil or 

Nutrient Capture 


the soil close to a tree. Root activity cannot be measured as an absolute 
amount of nutrient taken up per unit soil volume, but only as a so-called 
root activity distribution. The central assumption is that the tracer (for 
example, 3 -P) is taken up in all plant parts in the same proportion as the 
corresponding unlabelled nutrient (total phosphorus) (IAEA, 1975). 

Measurement of the vertical distribution of nutrient uptake 

A relative measure of the nutrient uptake from different soil depths by a 
plant can be obtained by applying a point source of a tracer at different 
depths through a tube. For each depth, the tracer has to be applied to 
different plant individuals, and relative nutrient uptake is calculated from 
the concentration of the tracer in tissue samples of these plants (Fig. 8.3). 
Several tubes should be installed for each plant to label the soil as 
homogeneously as possible and to raise the signal in the leaves to a level 
that can be conveniently measured. It is evident that a completely uniform 
labelling cannot be achieved with this method. Therefore, the total uptake 
per unit soil volume cannot be calculated, as mentioned above, but different 
depths can be compared and the uptake from one depth in relation to the 
uptake from all labelled depths can be calculated. If measurements are 

Tree species 

Foliar 32 P 
analysis k 

Tree 1 

Tree 2 

Tree 3 

32 P application 


0.1 m depth — tJ 

0.6 m depth 

Tree 1 
0.1 m 

1 .5 m depth 

Tree 2 
0.6 m 

Tree 3 
1 ,5 m 


Bactris gasipaes 

Theobroma grandiflorum 

(cpm g 1 ) 
(% of sum) 
(cpm g" 1 ) 
(% of sum) 
















Fig. 8.3. Schema of the application of a 32 P tracer at three depths to different tree 
individuals and calculation of relative uptake per depth for two Amazonian tree 
crop species. Radioactivity in leaf samples of the trees is given as cpm g~ 1 dry 
matter (after Lehmann et al., 2001b). 

176 G. Schroth and J. Lehmann 

carried out with individual trees, the tubes should be installed in a 
concentric ring around the tree, because uptake may vary between different 
distances from and sides of a tree (IAEA, 1975). The International Atomic 
Energy Agency (IAEA, 1975) arbitrarily chose 16 tubes per tree and found 
a high coefficient of variation (30-157%) for the root activity patterns of 
various tropical tree crops. Increasing the number of tubes to 60 per tree 
did not reduce this variability (with the exception of coconut palms). 

The following tracer materials can be used in nutrient uptake studies: 
stable isotopes such as 15 N, radioisotopes such as 32 P or 33 P, and rare 
elements such as strontium, lithium or rubidium. 


The use of radioisotopes in studies of root activity in tree-based production 
systems has recently been reviewed by Wahid (2001). In most of the 
published experiments for determining root activity patterns 32 P was used 
as a tracer. The tree species investigated included tropical and temperate 
fruit trees and fibre crops as well as leguminous service trees (Table 8.1). 
Phosphorus has two advantages over other tracers. First, it is relatively 
immobile in the soil and therefore remains at the point of application. 
Secondly, double labelling with 32 P and 33 P offers the possibility of 
comparing the phosphorus uptake from two different depths or distances 
on the same plant, thus considerably reducing the variability of the results 
compared with a design where the same tracer is applied at different 
depths or distances under different plant individuals (IAEA, 1975). 32 P is 
quite cheap but 33 P is expensive. Depending on the depth of application 
and nutrient uptake, 2-5 mCi should be used for one tree. The 
radioisotope should be applied together with about 5 mg of unlabelled 
phosphorus (as 1000 ppm P solution) to prevent a large part of the tracer 
phosphorus becoming fixed to oxides in highly weathered soils and hence 
unavailable to the target plant. Leaf sampling should be done as soon as 
an interpretable signal can be detected to avoid the decreasing pool size 
of the tracer affecting its uptake. The plant material is usually digested 
using heat and acid treatments, and the radioactivity is determined with 
a scintillation counter (IAEA, 1975). A disadvantage of radioactive 
materials is that they are a health hazard so transport regulations may 
seriously complicate their use. 

As mentioned above, a further disadvantage of tracer techniques with 
radioactive phosphorus is the high variability of the results. Coefficients 
of variation between replicate trees of 100% or higher for single-labelling 
experiments and of 50% or higher for double-labelling experiments are 
not uncommon (IAEA, 1975). The high variability of data obtained with 
the single-labelling approach is largely an effect of confounding variability 

Nutrient Capture 


Table 8.1. Maximum investigated root-activity 

distribution of seve 

ral tree species 

using 32 P. 






Cocoa (Theobroma cacao) 





Cocoa (Theobroma cacao) 




Wahid etal., 

Apple (Malus domestica) 





Cotton (Gossypium hirsutum) 




Bassett et 
al., 1970 

Mango (Mangifera indica) 




Bojappa and 
Singh, 1974 

Coffee (Coffea arabica) 




Huxley et al., 

Coffee (Coffea arabica) 




IAEA, 1975 

Coffee (Coffea arabica) 3 - 

Costa Rica 



Saiz del Rio 
etal., 1961 

Guava (Psidium gujava) 




Purohit and 

Gliricidia sepium 




Vasu et al., 

Orange (Citrus sinensis) 




IAEA, 1975 

Banana (Musa sp.) 




IAEA, 1975 

Citrus sp. 




IAEA, 1975 

Coconut (Cocos nucifera) 




IAEA, 1975 

Coconut (Cocos nucifera) 

Sri Lanka 



IAEA, 1975 

Oil palm (Elaeis guineensis) 




IAEA, 1975 

determined with radioactive 86 Rb. 

between tree individuals with variability in nutrient uptake from different 
soil depths, because uptake ratios between depths have to be calculated 
with data from different trees. This variability cannot be interpreted in a 
meaningful way. In contrast, the variability of the double-labelling method 
is caused by: (i) the heterogeneity of root-activity distribution between trees 
of the same species; and (ii) the spatial heterogeneity of the root activity 
of a tree at a given soil depth. Both of these are relevant research topics. 
For example, the spatial heterogeneity of tree root activity determines the 
'mesh size' of safety nets against nutrient leaching (see Section 7.1). 

178 G. Schroth and J. Lehmann 

Stable isotopes 

Stable isotopes like 15 N have only rarely been used for root-activity 
measurements of annual crops (Gass et al., 1971; Menezes et al, 1997) or 
trees (IAEA, 1975; Atkinson et al., 1978; Roweetal., 1999; Lehmann etal., 
2001b). The main obstacle for using 15 N is the mobility of nitrate, which is 
rapidly formed from the applied nitrogen source due to the high 
nitrification rates in many tropical soils. This may not be an important 
problem in acid soils with variable-charge clays and high oxide contents, 
which possess a high anion exchange capacity in the subsoil so that the 
mobility of nitrate is reduced (see Box 8.1). The mobility of the applied 
15 N in the soil can also be reduced by applying the tracer together with a 
labile carbon source like sucrose which provokes microbial growth and 
immobilization of the nitrogen near the point of application (Rowe et al., 
1999). The uptake of 15 N is very fast and leaf samples can be collected for 
analysis 2-4 weeks after the application. The root activity distribution of 
several tropical fruit trees measured with 15 N was very similar to that 
measured with 32 P (Lehmann et al., 200 lb), confirming earlier results from 
apple trees (Broeshart and Netsinghe, 1972). In the former study, 15 N was 
better suited for root activity measurements than 32 P, because it showed a 
lower variability between replicates, probably due to the labelling of a 
larger soil volume (Lehmann et al., 2001b). Preconditions for the successful 
use of 15 N in nutrient capture studies may include high microbial activity, 
high oxide contents, acid soil conditions and little leaching. Due to the 
large dilution of nitrogen in plant and soil, however, highly enriched 15 N 
sources have to be used, which are very expensive. The application of 
about 1-2 g 15 N in excess of natural abundance is recommended, 
preferably at 10 atom% 15 N excess or higher. The measurements are 
usually done by dry combustion coupled with mass spectrometry (Barrie 
et al., 1995), using samples as small as 5-15 mg. 

Rare elements 

An alternative to isotopic tracers is the use of rare elements. Strontium, 
which is chemically related to calcium and magnesium, is readily available, 
non-hazardous, inexpensive and easy to analyse by atomic absorption 
spectrometry. Furthermore, strontium is more stable in soil than nitrate 
and does not decay like the radioisotopes. The relation of strontium and 
calcium in the biomass of crop plants was similar to the relation of 
exchangeable strontium and calcium in the soil in which the plants were 
grown (Menzel and Heald, 1959). Fox and Lipps (1964) measured similar 
root-activity patterns of lucerne (Medicago sativa) with strontium and 32 P. 
Van Rees and Comerford (1986) measured the uptake of strontium by 

Nutrient Capture 179 

pines (Pinus elliottii) and understorey vegetation from different soil depths. 
The drawback of this method is its low sensitivity. Also, if large amounts 
of strontium have to be applied, it may cause nutrient uptake disorders 
and growth depression of the plants (Fox and Lipps, 1964). Rubidium and 
lithium are chemically related to potassium, but are rarely used as tracers 
because the former is expensive and the latter is phytotoxic. An exception 
is the work by Fitter (1986). 

Measurement of the lateral distribution of nutrient uptake 

For the measurement of lateral nutrient uptake, i.e. from different 
distances from the plant, the same techniques can be used as for the 
nutrient uptake from different depths using strip or point applications of 
32 P (IAEA, 1975; Wahid, 2001), 15 N (Atkinson et al., 1978) or strontium 
(Lehmann et al., 1999b). The tracers can either be applied at a certain 
depth or on the soil surface, and this offers the possibility of labelling whole 
areas (Lehmann et al., 2000). This approach allows the calculation of 
uptake per unit area, either relative to the amount of labelled fertilizer 
applied to this area or relative to a reference area. If the area to be labelled 
is large, stable isotopes are preferable to radioisotopes because of the lack 
of health hazards and over-rare elements because of the smaller amounts 
needed. If the above-ground biomass can be determined, the total 
recovery of the applied source can be calculated, which may be more useful 
information than root-activity comparisons between distances. Additionally, 
the relative importance of different areas for nutrient uptake can be 
compared, for example beneath the hedgerow and in the alley between 
hedges in hedgerow intercropping experiments (Lehmann et al., 2000). 

With all tracer applications, care has to be taken to maintain sufficient 
distance between treatments in order to avoid cross-contamination. This 
is especially true for radioisotopes because of the high sensitivity of the 

Tracer techniques for measuring root-activity distribution have certain 
advantages compared with root distribution studies and can, therefore, be 
a useful supplement to these. Tracers are only taken up by active roots, 
which are often difficult to distinguish visually from dormant or dead roots. 
They can also be used to separate nutrient uptake by different species from 
the same soil volume, whereas the visual distinction of roots from different 
species is often difficult (see Section 12.3). 

Chapter 9 

Nutrient Exchange with the Atmosphere 


1 Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil; 2 Institute of Agricultural Chemistry, University of Bonn, 
Karlrobert-Kreiten-Str. 13, 531 15 Bonn, Germany 

9.1 Synopsis 

Vegetation exchanges nutrients with the atmosphere through several 
processes, some of which can be influenced by agroforestry techniques. In 
this chapter, the effects of trees on nutrient gains through atmospheric 
deposition and nutrient losses through fire are outlined. Biological 
nitrogen fixation is another important atmospheric exchange, which is 
covered in Chapter 13. 

Nutrient gains from atmospheric deposition 

Atmospheric deposition of nutrients occurs either in a wet form associated 
with rain or fog, or in a dry form as particles or gases. Estimates of 
atmospheric inputs for different tropical sites can be found in Bruijnzeel 
(1991) and Szott et al. (1991), who concluded that annual inputs are in 
most cases low, although substantial quantities of nutrients may be 
deposited over long time spans. A critical review of atmospheric 
phosphorus inputs has been provided by Newman (1995), who points to 
methodological and other error sources in published values. Atmospheric 
nitrogen inputs for West Africa are given by Robertson and Rosswall (1986) 
as part of a regional nitrogen balance. For sub-Saharan Africa, Smaling et 
al. (1993) reported the following linear relationships between wet 
deposition of nitrogen (N), phosphorus (P) and potassium (K) in kg ha -' 
year 1 and the square root of average rainfall (r) in mm year- 1 : 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 181 

F.L. Sinclair) 

182 C. Schroth and J. Burkhardt 

N = 0.14r 05 (9.1) 

P = 0.023r° 5 (9.2) 

K = 0.092r° 5 (9.3) 

The presence of trees in agroforestry systems can influence 
atmospheric nutrient deposition by rain through the interception of 
rainfall and modification of the wind field (D.M. Smith et al., 1997a). Tree 
rows may increase rainfall and corresponding nutrient inputs on the 
windward side and reduce them on the leeward side (Wallace, 1996). Trees 
with high stemflow, such as certain palm species, concentrate atmospheric 
water and nutrient inputs in the proximity of their stem (Schroth et al, 
1999b, 2001a). 

Dew formation occurs mainly on leaf surfaces exposed to clear sky and 
is, therefore, influenced by trees through the amount of leaf area they 
support and the orientation and angular distribution of leaves. 
Condensing dew dissolves hygroscopic materials on the leaf surface that 
may have originated from dry-deposited aerosols (Burkhardt and Eiden, 
1990). The deposition of water-soluble gases such as ammonia or nitric 
acid is enhanced by dew (van Hove et al., 1989), although the stable 
atmospheric conditions generally occurring when dew is formed may limit 
the effectiveness of this mechanism. The chemical composition, especially 
the pH, of dew is determined by dissolved materials including ions leached 
from the leaf, and this may influence gas dissolution. Revolatilization and 
denitrification of dissolved components may occur as dew dries (Takenaka 
et al, 1999). 

In contrast to deposition of pure water as dew, fog is deposited as 
liquid water already including dissolved substances. Fog droplets are 
deposited by impaction and interception, and the leaves of trees often act 
as very effective fog collectors, especially in mountainous regions where 
clouds are directed by wind towards hills (cloud-stripping; Doumenge et 
al., 1995). In some cloud-forest areas, fog interception is an important 
water source (Cavelier, 1996) and would also be expected to contribute 
significantly to atmospheric nutrient deposition. In the northern Andes, 
a cloud belt is located at 1000-1400 m. In other regions, the altitude of 
such zones of elevated atmospheric humidity varies with climatic and 
geographical factors, such as general humidity and distance to the sea 
(Cavelier, 1996). 

Dry deposition includes nutrients in dust and aerosols. These may be 
transported over long distances, such as from the Sahara to the West 
African savanna and rainforest zones by the Harmattan winds, or may 
come from local sources, such as soil and vegetation. In Niger, average 
dust inputs from the Harmattan of 1500 kg ha _1 year 1 , containing 2.7 kg 
of calcium, 0.9 kg of magnesium and 1.2 kg of potassium, were measured 

Nutrient Exchange with the Atmosphere 183 

with open-bucket samplers. The dust contained 10-30 times more 
exchangeable calcium, magnesium and potassium than the soil on which 
it was deposited, and the cation exchange sites of the dust particles were 
base saturated, whereas the topsoil was only 20-30% base saturated. The 
dust input could thus have a significant effect on soil chemistry (TropSoils, 
1989). With increasing distance from the Sahara, the annual dust depo- 
sition generally decreases; however, a deposition of 2.5 kg ha _1 year 1 of 
potassium and 3.5 kg ha -1 year ' of calcium was still measured in the Ta'i 
forest in the south-western Cote d'lvoire in another study (Stoorvogel et 
al., 1997). This dust is low in nitrogen due to its Saharan origin (Robertson 
and Rosswall, 1986). Plant pollen is a source of dust that is relatively rich 
in phosphorus (Newman, 1995). 

The presence of trees strongly increases the deposition of dust and 
aerosols by filtering them from the air with their crowns (Stoorvogel et al. , 
1997). Deposition of dust on trees, from which it is washed down by 
subsequent rains, could contribute to higher clay and nutrient contents in 
the soil under savanna trees (Sanchez, 1987; Young, 1997), although there 
are numerous alternative explanations, such as preferential establishment 
of trees on more fertile sites, vertical and lateral redistribution of nutrients 
by tree roots, animal droppings under trees, and differences in soil erosion 
between areas with and without tree cover. Wind deposition may be 
uneven over short distances and contribute to soil heterogeneity. Adderley 
et al. (1997), for example, reported surface sand derived from lacustrine 
and aeolian deposits, varying from to 690 mm in depth over clay soils 
across a 30 ha site in northern Nigeria, with a mean change of 0.46 mm 
m- 1 from north to south attributable to airborne deposition. Tree survival 
at this site was highly correlated with the sand content of surface soil. 

Effects on the wind field and thus on rain distribution and dust 
interception are greatest for large trees. Pronounced effects would be 
expected from windbreaks, boundary plantings and perhaps from large 
solitary trees in crop fields, pastures or shaded tree crop plantations. Small, 
regularly pruned trees would not be expected to have significant effects 
on atmospheric deposition (Szott et al, 1991). 

Nutrient losses with fire 

Agroforestry practices may also affect carbon and nutrient exchange with 
the atmosphere by reducing the frequency of fires. Fire is a common 
management tool in traditional agriculture in both forest and savanna 
ecosystems. It serves for site clearing, transformation of biomass into 
nutrient-rich ash, weed and pest control, pasture regeneration and 
hunting. Burning early in the dry season can prevent much more intensive 
and destructive fires in the late dry season. In simultaneous agroforestry 

184 C. Schroth and J. Burkhardt 

practices, the presence of the trees may be an important incentive for 
farmers to protect their land from spontaneous fires, especially in savanna 
areas. Belts of fire-resistant trees under which the ground vegetation is 
suppressed and litter periodically burned are an agroforestry technique 
that can help to prevent the spread of bush fires. Improved fallows should 
also not be burned before cropping, especially when the nitrogen 
accumulation in legume biomass is an objective of the fallowing, although 
farmers may prefer to burn to facilitate site clearing (MacDicken, 1991). 

During fires, nutrient losses to the atmosphere occur through 
volatilization. In addition, ash particles are carried away by wind and the 
chimney effect generated by the fire (Mackensen et al., 1996). Fire accounts 
for the bulk of all nitrogen losses from West Africa (Robertson and 
Rosswall, 1986), and empirical evidence suggests that ecosystems affected 
by recurrent burning as occurs in savannas are characterized by relatively 
low nutrient stocks (Ehrlich et al., 1997). In addition to the nutrient losses, 
the release of the greenhouse gases carbon dioxide, carbon monoxide, 
methane and nitrogen oxides as well as aerosols during the combustion 
process has drawn the attention of researchers concerned about climate 
change to vegetation fires in the tropics (Carvalho et al., 1995; Kauffman 
etal, 1998). 

During the burning of two secondary forests and one fallow site in 
eastern Amazonia, Mackensen et al. (1996) measured the following element 
losses: 95-98% of the nitrogen in the biomass, 67-76% of the sulphur, 
27-47% of the phosphorus, 16-48% of the potassium, 17-43% of the 
magnesium, 9-35% of the calcium and 17-30% of the sodium, together 
with 94-98% of the carbon. The stems were removed from the forest sites 
before the burn, leaving only relatively small material, which explains the 
high carbon and nitrogen losses. Where tree stems are burned, lower 
combustion efficiencies are achieved (Carvalho et al., 1995; Kauffman et 
al. , 1998). When burning secondary forests in Brazilian Amazonia, Hughes 
et al. (2000) found that the average proportions of the nitrogen, sulphur, 
phosphorus, potassium and calcium in the above-ground biomass that 
were lost from the site through fire were 70, 54, 33, 17 and 20% of the pre- 
fire pools. Kauffman et al. (1998) identified three factors influencing 
nutrient losses from vegetation fires: (i) the level of biomass consumption 
(nutrient losses increase with increasing combustion efficiency); (ii) 
nutrient distribution within biomass classes with varying susceptibility to 
combustion losses (lower nutrient losses from stems than from fine litter); 
and (iii) temperatures of volatilization of different elements (greater losses 
for nitrogen than for phosphorus, potassium and calcium). Effects of fire 
on soil properties have recently been reviewed by Neary et al. (1999). 

Nutrient Exchange with the Atmosphere 185 

9.2 Methods for Atmospheric Nutrient Inputs 

Wet deposition of nutrients at a site can be approximated by collecting 
rainfall in an open area or above the canopy with funnels and multiplying 
the periodical rain totals with the corresponding nutrient concentrations 
(Bruijnzeel, 1991). Bias can result from strong winds during rainfall 
leading to edge effects which affect collector yields, or from dry-deposited 
material in the collectors including dust, vegetation debris or insects. 
Coarse material may be prevented from entering the collectors by grids 
or ping-pong balls. If funnels are used, adhesion of water to the surface 
may lead to underestimation of water inputs, especially for small rains. 

The determination of dry deposition to an ecosystem poses more 
difficulties, as this term summarizes a wide range of particle size classes 
from gas molecules to dust particles, extending over five orders of 
magnitude and therefore involving considerably different deposition 
mechanisms. Whereas for dust particles (>10 Lim in diameter) 
gravitational settling is dominant, the deposition of smaller aerosol entities 
is dependent on characteristics of the acceptor, such as geometry, 
microroughness and chemical composition. For the latter group, artificial 
samplers will not give realistic results in most cases. Four approaches to 
the determination of dry deposition are touched upon below. 

Micrometeorological methods 

Micrometeorological measurements such as eddy correlation are most 
effectively used for gases. They require very specialized and cost-intensive 
equipment and mostly have homogeneous terrain as a precondition, which 
makes them unsuitable for heterogeneous agroforestry conditions 
(Finkelstein and Sims, 2001). 

Washing of plant surfaces 

Leaves, twigs or whole trees have been washed with water to quantify 
atmospheric inputs (Reiners and Olson, 1984). Washing procedures of 
leaves may answer specialized questions regarding aerosol and gas 
deposition, but the dependence of leaf orientation and microsite within 
the canopy may be decisive for the amount of deposited material, so 
extensive data sets are needed to obtain a complete picture. Furthermore, 
the efficiency of washing depends on the age of the leaf, and there are 
effects of leaching, which will mostly be dependent on the washing time. 
Leaf washing may be especially useful if leaves of different, neighbouring 

C. Schroth and I. Burkhardt 

trees are compared, which may have different sampling characteristics for 
dry-deposited material. 

Artificial filters 

Artificial filters for dry deposition of nutrients can be calibrated against 
natural foliage and exposed in different positions of tree canopies 
(Kellman and Carty, 1986). As in the method discussed before, the effort 
for measuring inputs in different crown positions may be considerable. 

The canopy balance method 

This method has been described by Ulrich (1983) and Lovett and Lindberg 
(1984). It consists of the comparison between rainfall, which is sampled 
above the canopy or on a nearby location without trees, and throughfall 
and stemflow, sampled within the canopy, assuming that these contain the 
substances that were dry deposited on the canopy before the respective 
rain event. Much less equipment than in micrometeorological 
measurements is needed and aerodynamically homogeneous conditions 
are not required to obtain a time-integrated overall picture of dry 

However, nutrient deposition in throughfall and stemflow under trees 
would correspond to total atmospheric deposition only if it could be 
assumed that: (i) uptake of deposited nutrients through the leaves is 
negligible in comparison to the deposited quantities; and (ii) there is 
negligible leaching of root-absorbed nutrients from the canopy (Cape, 
1993). These assumptions are usually not correct. Besides the nutrients 
derived from atmospheric deposition, throughfall and stemflow contain 
nutrients that were taken up by the trees from the soil and were leached 
by the rainwater from foliage and bark. This latter part reflects mere 
recycling of nutrients, similar to much of the nutrients in litterfall. 
Furthermore, certain elements such as hydrogen and ammonium ions are 
taken up from throughfall water by the plants during crown passage and 
are replaced by other cations (especially calcium, magnesium, potassium 
and manganese), which are leached from the leaves and therefore increase 
in the throughfall water (Tukey, 1970; Lovett, 1992). The recent detection 
of microscopic waterfilms extending from the leaf surface into the stomata, 
through which nutrient ions may move into the leaf, indicates that foliar 
nutrient uptake may even continue during prolonged periods of the day 
when no free water is present on the leaves (Burkhardt and Eiden, 1994; 
Eichert et al., 1998). In addition, epiphytes may retain nutrients from 

Nutrient Exchange with the Atmosphere 187 

atmospheric deposition and fix nitrogen, and these nutrients may be 
released in pulses following drying cycles (Coxson, 1991). 

The canopy balance method therefore requires the separation of those 
nutrients in throughfall and stemflow that originate from atmospheric 
deposition from those that were taken up by the plant roots and were 
leached from the foliage. For this, sodium has been used as a natural 
tracer. Sodium is not leached from the surface of most plants in significant 
quantities, and it can thus be assumed that all the sodium on a plant surface 
comes from the atmosphere (Newman, 1995). Kellman and Carty (1986) 
measured dry deposition of nutrients in pine trees with artificial filters 
which were exposed in the canopy. The deposition of sodium on the filters 
was calibrated against that on natural foliage by exposing the filters 
together with prewashed pine whorls to the atmosphere and analysing 
leachates from filter and foliage for deposited nutrients. It was assumed 
that other elements were deposited at similar ratios on filters and foliage 
to sodium. The nutrient deposition on the trees could thus be calculated 
from that on the filters (Kellman and Carty, 1986). 

Instead of exposing aerosol filters at different positions in a tree 
canopy, sodium can also be measured in throughfall and stemflow from 
the respective trees. With the ratio of sodium and other elements measured 
in dust filters, and after subtracting the sodium input from wet deposition 
(rainwater), the dry deposition of different elements on the tree canopy 
can then be estimated (Ulrich, 1983). In both cases, the assumption that 
other elements behave in a similar way to sodium with respect to their 
deposition on artificial filters and foliage is open to doubt, as different 
elements may be contained in aerosols and dust particles differing widely 
in size and surface properties (Sehmel, 1980; Newman, 1995). The 
methods described here can thus only provide an approximation of dry 
deposition. Apart from sodium, titanium has also been used as reference 
element assumed to have purely atmospheric origin (Stoorvogel et al., 
1997). The separation of atmospheric deposition from canopy leaching as 
a component in throughfall and stemflow has also been attempted with 
radioactive tracers applied to the soil (Cape, 1993). 

The heterogeneity of throughfall may be considerable, and large 
numbers of samplers may be necessary to obtain a representative average 
value for an area (Draaijers et al, 1996). To improve the precision of the 
estimate and/or reduce the number of collectors required, their positions 
can be changed from time to time (Lloyd and Marques, 1988). However, 
the spatial heterogeneity of nutrient inputs in throughfall and stemflow, 
as affected by agroforestry trees with different crown architectures, may 
also be of considerable interest itself (Schroth et al., 2001a). Gaseous 
deposition of nutrients to the stomata is not included in measurements 
using the canopy balance method. 

C. Schroth and I. Burkhardt 

Sample collection and analysis 

Rainwater, throughfall and stemflow are usually very dilute solutions that 
are highly susceptible to contamination both in the field, from leaves, 
insects, birds, mineral fertilizer, etc., and in the laboratory. Thus, collectors 
should be washed very carefully, and frequent collection of the samples is 
advisable to reduce the risk of sample loss by contamination and to prevent 
chemical alterations due to microbial action, affecting especially phos- 
phorus and ammonium. Stemflow samples often contain organic particles 
and need to be filtered before analysis. Filtration at 0.45 pm is sometimes 
used to produce normalized conditions and reduce microbial activity, but 
filtration with prewashed paper filters is also acceptable. The inclusion of 
organic nitrogen and phosphorus forms in the analyses is essential, as these 
may contribute significantly to the total inputs of these nutrients (Schroth 
et al, 2001a). 

9.3 Methods for Nutrient Losses from Burning 

According to Mackensen et al. (1996), three approaches to measuring 
nutrient losses during biomass burning can be distinguished: (i) collection 
of smoke samples above the fire, which are analysed for gases and ash 
particles; (ii) simulation of the fire in a muffle furnace (where flame 
conditions differ from those in the field and losses of ash particles cannot 
be simulated); and (iii) quantification of carbon and nutrient stocks in the 
biomass and ash before and after the burn in the field. Only the latter 
approach will be discussed here, as it is probably the most useful in 
agroforestry research. Losses to the atmosphere in gaseous and particulate 
form cannot be measured separately with this approach, but this may often 
not be necessary. 

The method requires the estimation of the mass (often calculated from 
measurements of volume and density) and nutrient concentration of the 
biomass before and after the burn. The volume of stems and woody debris 
on the ground can be estimated non-destructively with the line intersect 
(or planar intersect) method (van Wagner, 1968; Brown and 
Roussopoulos, 1974). The wood particles are divided into diameter classes, 
and the intersections of the particles with sampling lines are counted. The 
total volume V of the biomass is calculated as 

V = na [(n 2 d q 2 )/8L] (9.4) 

where n is the number of intersections, d q is the average squared diameter 
of the particles in a diameter class, L is the length of the sampling line, and 
a is the correction for non-horizontal orientation of wood particles and is 
defined as the average secant (= cos 1 ) of the angle of the particles with 

Nutrient Exchange with the Atmosphere 

the horizontal. Sampling bias due to non-random orientation of the wood 
can be minimized by using several lines placed at different directions. In 
an Amazonian pasture, Kauffman et al. (1998) used 32 systematically 
distributed sampling lines both before and after a burn. Wood particles 
were divided into the diameter classes 0-0.64 cm, 0.65-2.54 cm, 2.55-7.5 
cm, 7.6-20.5 cm and >20.5 cm, and increasing lengths of the sampling 
lines were used for larger diameter classes (1 m for the smallest and 15 m 
for the largest class). Average angle and diameter were measured on 100 
randomly selected particles of each size class. Samples for wood density 
and nutrient concentration were also collected per size class. 

The preburn nutrient stocks have to be determined shortly before the 
burn, because nutrients may be leached from the biomass into the soil 
when the cut vegetation is exposed to rain before the fire, and this reduces 
potential losses to the atmosphere. After the burn, the nutrient stocks are 
measured both in the unburned residues and in the ash and charcoal. 
Again, measurements have to be taken immediately, as nutrients may be 
lost from the ash by leaching into the soil without necessarily being lost 
from the site, or by particle erosion. For the quantitative collection of ash, 
steel trays may be placed on the soil under the vegetation debris 
(Mackensen et al, 1996), or the ash may be collected from microplots, e.g. 
with a vacuum cleaner (Kauffman et al., 1998). 

Chapter 1 
Soil Structure 

M. Grimaldi, 1 G. Schroth, 2 W.G. Teixeira 3 and B. Huwe 4 

' Institut de Recherche pour le Developpement (IRD), UR 137, Centre 
de Recherche d'fle de France, 32 ave Henri Varagnat, 93143 Bondy 
Cedex, France; 2 Biological Dynamics of Forest Fragments Project, 
National Institute for Research in the Amazon (INPA), CP 478, 
6901 1-970 Manaus, AM, Brazil; 3 Empresa Brasileira de Pesquisa 
Agropecuaria-Amazonia Ocidental, CP 319, 69011-970 Manaus, 
AM, Brazil; 4 lnstitute of Soil Science and Soil Geography, University 
of Bayreuth, 95440 Bayreuth, Germany 

10.1 Synopsis (C. Schroth) 

Soil structure comprises the arrangement of individual particles in the soil 
and is especially concerned with their aggregation, the resulting pore size 
distribution and the stability of the aggregated state (Payne, 1988). Soil 
structure affects the availability and mobility of water, air and nutrients 
and influences the growth of plant roots, which is generally reduced in 
compact, dry and poorly aerated soil. By controlling the access of plants 
to water and nutrients, soil structure determines, to a large extent, the 
productivity of a site. The maintenance of an adequate soil structure is an 
essential element of the sustainability of land-use systems. Where land-use 
practices have led to structural degradation of the soil, the improvement 
of soil structure is often a precondition for the regeneration of site 
productivity. This is usually much more difficult to achieve than the 
correction of chemical fertility problems (Alegre et ah, 1986). 

Structural degradation of agricultural soils 

Soil structural degradation typically occurs when forest or savanna 
vegetation is replaced by annual cropping systems. Under continuous 
cropping with maize for 3 years on a coarse-textured Alfisol in the Nigerian 
rainforest zone, increased bulk density, reduced soil porosity, reduced 
water infiltration, surface crusting and reduced water-holding capacity 
compared with fallow soil indicated the progressive degradation of soil 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 191 

F.L. Sinclair) 

192 M. Crimaldi etal. 

structure. The degradation was apparently caused by severe loss of soil 
organic matter, surface erosion and reduced earthworm activity under the 
annual crops (Juo and Lai, 1977). Similarly, annual cropping in the 
wooded savanna zone of West Africa has often led to increased bulk density 
and reductions in porosity, structural stability and water infiltration (Pieri, 
1989). Cropped ferralitic soils in Queensland, Australia, have developed 
an increased tendency for surface sealing under high-intensity rainfall, 
associated with a decline in soil organic matter (Bell et al., 1998). 
Degradation of soil structure has also been observed under tree crops such 
as cocoa (Theobroma cacao; Ekanade, 1985). 

Soil structural degradation often begins when inappropriate 
land-clearing methods are used. On an Oxisol in central Amazonia, 
mechanical forest clearing destroyed many pores with an equivalent size 
of >0. 1 (im in the topsoil, thereby reducing the amount of plant-available 
water (Chauvel et al., 1991). Physical soil degradation caused by 
mechanized land clearing has also been described for an Ultisol in the 
Peruvian Amazon (Alegre et al., 1986) and an Alfisol in West Africa 
(Hulugalle, 1994) and contributed to yield losses in the second rotation of 
an oil palm (Elaeis guineensis) plantation on a ferralitic soil in the Cote 
d'lvoire (Caliman et al., 1987). 

Potential of trees to improve the structure of agricultural soils 

Soil structure is stabilized by the presence of binding materials such as clay, 
if flocculated by polyvalent cations such as calcium or aluminium, lime, 
and oxides and oxihydroxides of iron and aluminium (Payne, 1988). The 
action of these binding materials is not influenced by the introduction of 
trees to agricultural systems and so will not be further discussed here (see 
also Section 10.3). However, the potential of trees to contribute to the 
maintenance and rehabilitation of soil structural characteristics in other 
ways has been well established. Soil physical characteristics have often been 
improved in hedgerow intercropping plots compared with control plots 
with annual crops (Rao et al., 1998). The form of these improvements has 
included better soil aggregation, lower bulk density, lower resistance to 
penetration, reduced surface sealing, improved soil porosity and 
consequently higher water infiltration and water-holding capacity. Similar 
positive effects have been found under planted fallows. The integration of 
trees in agricultural systems affects soil structure through the following 

• Trees may increase the quantity of litter or mulch on the soil surface, 
which prevent rain drops hitting and breaking down soil aggregates, 
which leads to clogging of infiltration paths. Reduced overheating and 
evaporation from the soil surface are added advantages of a litter layer. 

Soil Structure 1 93 

In the Nigerian rainforest zone, negative effects of annual cropping 
on soil structure were much reduced when the soil was mulched, as 
this avoided crusting and maintained a higher earthworm activity (Juo 
and Lai, 1977; Juo etal., 1995; see also Chapter 16). 

• Trees can increase soil organic matter levels and microbial activity 
through increased inputs of above- and below-ground biomass, 
reduced soil temperature and reduced erosion (see Chapters 4 and 
17). Soil organic matter has a stabilizing effect on soil aggregates, and 
the loss of humus associated with cultivation is usually accompanied 
by structural degradation. Addition of easily decomposable organic 
residues to soils, such as certain tree prunings, leads to the synthesis 
of polysaccharides and other organic compounds that stabilize soil 
structure (Stevenson and Cole, 1999). Hyphae of saprophytic and 
mycorrhizal fungi can bind mineral and organic particles together into 
stable aggregates (Tisdall et al. , 1997), and part of the stabilizing effect 
of plant roots on the soil structure is due to their association with 
mycorrhizas (Miller and Jastrow, 1990). Microbial polysaccharides 
produced in the rhizosphere could also contribute to this effect 
(Angers and Caron, 1998). 

• The introduction of trees to agricultural systems will influence the 
quantity and types of roots present in the soil, which may affect soil 
structure. Plant root systems stabilize soil structure by enmeshing 
aggregates, releasing binding materials (mucilage) into the 
rhizosphere (Morel et al., 1991) and increasing soil microbial activity. 
As a result, root length or mass may correlate significantly with soil 
aggregation (Miller and Jastrow, 1990). When roots grow through the 
soil, they enlarge existing pores and create new ones, but their radial 
pressure may, at the same time, compress the soil around these pores. 
Water uptake by roots also affects the formation and fragmentation of 
aggregates by influencing wetting-drying cycles in the surrounding 
soil (Angers and Caron, 1998). The relatively large, continuous 
macropores resulting from root growth can play an important role in 
soil aeration and rapid macropore flow of water through the soil (see 
Chapter 11). In addition, old root channels are often used by new 
roots, which thereby avoid the mechanical resistance of the soil matrix 
and may profit from the more favourable chemical environment 
provided by root debris (van Noordwijk et al, 1991a). Plants differ in 
their ability to penetrate compact soil, and the increased soil porosity 
caused by plants with strong root systems, including certain tree 
species, could improve the root development of associated or 
subsequent crops with weaker root systems and thereby their access 
to additional soil water and nutrients. Like other macropores, such 
biopores could be especially important in facilitating root penetration 
when the soil matrix is too dry or its strength too high to allow rapid 

194 M. Crimaldi et al. 

root growth. This can be essential for the establishment of seedlings 
(Cornish, 1993). This biological drilling by plant roots (and soil fauna) 
is most important at sites with compact subsoil horizons, and trees and 
woody shrubs are especially effective because of their perennial nature 
and their known ability to penetrate very hard horizons. However, in 
their critical review of the subject, Cresswell and Kirkegaard (1995) 
point out that there is not always a clear causal relationship between 
crop yield increases in rotations with deep-rooting species (both 
herbaceous and woody) and the creation of macropores in compact 
subsoil. To demonstrate biological drilling as the reason for yield 
improvements under these conditions, it would be necessary to 
separate effects of improved subsoil structure from those of altered 
soil chemical properties or the carry-over of root diseases. In future 
studies of the topic, particular attention should be given to the 
hydrological conditions of roots growing in existing macropores as 
influenced by possibly incomplete root-soil contact (e.g. small crop 
roots growing in wide tree root channels) as well as the possible 
concentration of pathogenic microorganisms and nematodes in such 
biopores (Cresswell and Kirkegaard, 1995). Methods of studying 
interactions between roots and soil structure have been reviewed by 
van Noordwijk et al. (1993b). 
• Through the quantity and quality of root and shoot litter, 
microclimatic effects and avoidance of soil tillage trees may influence 
the abundance, composition and activity of the soil fauna, such as 
earthworms, termites and ants, which in turn may affect soil structure 
through their burrowing activity. Faunal effects on soil structure are 
further discussed in Chapter 16. 

These strongly interdependent mechanisms deserve the attention of 
agroforestry researchers because they provide a basis for the use of trees 
in the management of soil structure. A quantitative understanding of these 
mechanisms is necessary for the identification of criteria for agroforestry 
tree species that are efficient in the maintenance and regeneration of soil 
structure. On a sodic soil in India, improved soil physical conditions under 
Prosopis julifl,ora compared to Acacia nilotica was explained by the higher 
litterfall and, more importantly, the larger root system of P. juliflora (Garg 
and Jain, 1992). A comparison of nine leguminous tree species and a 
spontaneous control in central Cote d'lvoire provided indications that 
fallow species with high root mass are more efficient in the rehabilitation 
of compacted soils than species with lower root mass (Schroth et al. , 1996). 
Afforestation of grassland in the Philippines with Acacia auriculiformis 
improved topsoil physical characteristics and water infiltration, whereas 
Pinus kesiya did not have this favourable effect, apparently because of the 
high faunal activity and well-developed root system of A. auriculiformis as 

Soil Structure 1 95 

opposed to the dense fungal mycelia in the soil associated with P. kesiya 
(Ohta, 1990). More mechanistic studies relating tree characteristics to their 
effect on soil structure are clearly desirable. 

10.2 Methods for Soil Bulk Density (W.G. Teixeira, 
B. Huwe) 

Bulk density is a simple measure of soil structure. It is defined as the ratio 
of the mass of an oven-dry soil sample (dried at 105°C to constant weight) 
to its bulk volume. It is a temporally and spatially variable soil property 
that can be used as an indicator of changes in soil structure caused by 
agricultural management, root growth and activity of soil flora and fauna. 
In this section, only field sampling procedures will be discussed. For 
detailed laboratory procedures see Blake and Hartge (1986) and Mclntyre 
and Loveday (1974). 

Examples of agroforestry studies in which different plant species and 
management practices caused significant changes of bulk densities in 
topsoil and subsoil can be found in Torquebiau and Kwesiga (1996), Mapa 
and Gunasena (1995) and Huxley et al. (1994). A reduction of soil bulk 
density is frequently interpreted as an improvement of soil physical 
properties, which is in most cases appropriate. Nevertheless, a reduction 
in bulk density (and increase in porosity) can sometimes be 
disadvantageous, for example when increased water infiltration and 
percolation cause increased nutrient leaching. Also, the plant-available 
water capacity of coarse-textured soils may be lower at low than at high 
bulk density, and can then be increased by compacting the soil (Archer 
and Smith, 1972). 

Core method 

The soil sample is collected by driving a steel cylinder into the soil to the 
desired depth. The cylinder is then carefully removed to obtain an exact 
volumetric sample. Sampling is normally carried out by hammering or 
jacking. Especially for soils with large silt and clay content it is important 
to use appropriate samplers to avoid compaction of the sample (Mclntyre 
and Loveday, 1974). The core diameter should be at least 7.5 cm and 
preferably 10 cm if accuracy is critical, and the core length should ideally 
be equal to the diameter (Mclntyre, 1974). Sampling should be carried 
out at intermediate soil moisture conditions because sampling during very 
wet conditions may cause compression of the sample, and sampling during 
very dry conditions may result in shattering of the cores. In swelling soils, 

196 M. Crimaldi etal. 

bulk density depends on soil moisture, and bulk density data should thus 
either be accompanied by moisture data at the time of sampling or, 
preferably, be determined at a specified matric suction. A suitable moisture 
content for measuring bulk density in such situations is when lateral 
swelling has reached a maximum and all cracks are closed (Mclntyre, 
1974). If the aim of the sampling is only the measurement of bulk density, 
it is not necessary to keep the collected samples undisturbed or to avoid 
soil moisture loss. 

Clod method 

In situations where the sampling of cores is not practicable, as in gravelly 
soils, bulk density can be measured using soil clods. For such soils it is 
generally better to make measurements on a large clod with small 
disturbance than on several small cores that may have undergone serious 
disturbance during collection (Mclntyre, 1974). The two available 
techniques for measuring the bulk density of soil clods, paraffin wax 
coating and kerosene saturation, are only suitable for relatively stable clods. 
These techniques apply Archimedes' principle to determine clod volume 
by weighing the clod in air and in a liquid of known density. Particularly 
for expansive soils, this method can lead to an underestimation of bulk 
density due to expansion of the clod when restraining forces caused by the 
overlying soil are removed (Blake and Hartge, 1986). Expansive soils must 
therefore be equilibrated with a standard matric suction for comparison 
of their densities (e.g. 100 hPa; Mclntyre and Loveday, 1974). 

Other methods 

Field methods suitable for gravelly soils involve the excavation of a quantity 
of soil for drying and weighing and the determination of the volume by 
filling the hole with sand or placing a rubber balloon in it which can then 
be filled with water (Blake and Hartge, 1986; Liu and Evett, 1990). Nuclear 
techniques are also available but their use is common only in engineering 

10.3 Methods for Aggregate Stability (M. Grimaldi) 

The aggregate stability of soil affects infiltration of water and susceptibility 
to water erosion, crust formation, hardsetting and compaction (Angers 
and Mehuys, 1993). A large number of methods for evaluating the 
aggregate stability of soils have been proposed since the pioneering work 

Soil Structure 1 97 

of Yoder (1936). These testify, on the one hand, to the importance of this 
soil characteristic, and demonstrate, on the other hand, the difficulty of 
defining a universally applicable method. This difficulty arises because of 
the diversity of factors and mechanisms of disaggregation which act on 
different organizational levels of soil structure, and which should be 
reflected by the measurement. The principal mechanisms of disaggrega- 
tion are slaking of aggregates by the compression of entrapped air, 
disaggregation by differential swelling of clays which provokes a micro- 
fissuration of the aggregates, disaggregation by the impact of rain drops 
and physicochemical dispersion. 

The stability of aggregates depends on their mineral and organic 
constituents (Emerson and Greenland, 1990). Three types of constituents 
are particularly important: the percentage of exchangeable sodium 
(Shainberg, 1992), the oxides and oxihydroxides of iron and aluminium 
(Le Bissonnais and Singer, 1993), and soil organic matter (Chenu, 1989; 
Haynes and Swift, 1990). Of these, it is mainly the dynamics and 
distribution of soil organic matter that can be influenced by trees (see 
Section 10.1). 

A method of measuring aggregate stability that has recently been 
proposed by Le Bissonais (1996) integrates key aspects of the older 
methods while being adapted to a wide range of different soil types. It 
consists of three main treatments which simulate the three principal 
mechanisms of aggregate fragmentation: 

• treatment 1 : disaggregation by slaking, provoked by rapid immersion 
in water; 

• treatment 2: disaggregation by microfissuration, provoked by slow 
capillary wetting; 

• treatment 3: mechanical disaggregation, provoked by shaking in water 
after slow capillary wetting. 

The initial physical conditions of the samples, and especially their 
moisture and the size of the aggregates, have an important influence on 
the results of the treatments. These conditions are therefore standardized 
to allow the comparison between different samples, e.g. for different 
pedological horizons, climatic conditions or cropping systems. The specific 
soil characteristics such as their water content at the time of sampling and 
the plot history are taken into consideration when interpreting the results. 

The soil samples are sieved to obtain aggregates of a size between 3 
and 5 mm, followed by air-drying or equilibration at a certain matric 
potential. For treatment 1, the aggregates are immersed for 10 min in 
distilled water. This treatment can also be used as a qualitative, simple and 
rapid field test. For treatment 2, the aggregates are placed on a filter paper 
that is wetted with distilled water, and for treatment 3 they are wetted with 
ethanol and then manually shaken in distilled water. Wetting in ethanol 

M. Crimaldi ef al. 

reduces the slaking of the aggregates by reducing the speed of the water 
penetration and the swelling of the clay minerals. For each treatment, and 
thus for each of the three mechanisms of disaggregation, the distribution 
of the aggregates in seven size classes is determined (>2 mm, 
2-1 mm, 1-0.5 mm, 0.5-0.2 mm, 0.2-0.1 mm, 0.1-0.05 mm and <0.05 
mm). The detailed protocol is given by Le Bissonais (1996), who proposes 
synthesizing the results of the measurement by calculating the weighted 
mean diameter of the aggregates, which is the sum over all size classes of 
the product of the dry weight of the aggregates and the mean between the 
two limits of the respective size class. 

If physicochemical dispersion is the main mechanism of dis- 
aggregation in a soil, the three treatments lead to the same distribution of 
aggregate sizes with a large fraction <0.05 mm. In this case it is 
recommended to evaluate the quantity of dispersed clay in the suspension 
that is produced by one of the treatments. If, on the other hand, the 
disaggregating effect of all the three treatments is small, a treatment 
proposed by Henin et al. (1958) may still reveal differences between 
samples. It consists of wetting the aggregates in benzene before the 
immersion in water. Benzene increases the wettability of the pore walls 
and can therefore reveal the hydrophobic effect of certain organic 

An approach to evaluating the causal relationships between aggregate 
stability and correlated soil characteristics such as root length or length of 
fungal hyphae in a sample has been described by Jastrow and Miller 
(1991). These authors also provide a very useful discussion of the 
measurement of aggregate stability as related to biological soil processes. 

10.4 Methods for Soil Porosity and Pore Size Distribution 

(M. Grimaldi) 

The measurement of the parameters of the pore space of a soil, such as 
total porosity, pore size distribution and morphology of the pores, is a 
quantitative, indirect approach to the analysis of soil structure. The soil 
pores are discontinuities that affect the spatial distribution of mineral and 
organic soil constituents and allow us to distinguish between successive 
levels of organization of these constituents within the soil: clay domains or 
microaggregates, aggregates, clods and horizons. It is always important, 
but especially so in agroforestry situations, to consider the lateral variations 
of soil porosity as much as its vertical variations in the soil profile. Within 
a single soil horizon, aggregates or soil volumes differing widely in porosity 
can coexist, depending on the spatial patterns of faunal activity, root 
distribution or tillage practices. 

Soil Structure 199 

Soil porosity 

The porosity of a soil sample (n) is the volume of pores related to the total 
volume of the sample. It is calculated from bulk density (yd) and particle 
density (ys), both measured in g cm- 3 : 

n = \-^- (10.1) 

For the measurement of bulk density yd see Section 10.2. The particle 
density ys can be measured with submersion or pycnometer methods 
(Blake and Hartge, 1986). In mineral soils it is often assumed to be 2.65 
g cm 3 , the value for quartz and also kaolinite. However, this assumption 
can lead to errors for soils with high contents of organic matter (which 
reduces ys) or iron (which increases ys). For example, for a ferralitic soil in 
Amazonia, ys was 2.64 g cm- 3 with 0.8% organic carbon and 2.52 g cm-' 
with 4.3% of organic carbon. High iron contents - recognizable by the red 
soil colour - can increase ys to values close to 3 g cm- 3 . 

If the inspection of the soil profile reveals several more or less 
homogeneous units, such as horizons or subhorizons, samples for porosity 
measurements should be collected from each of these units. The necessary 
number of replicate samples for porosity measurements (as for any other 
measurement) depends on the required precision and variability of the 
porosity between individual samples (see Section 3.2), which vary widely 
between different field situations. Although it is not possible to define a 
priori the required number of replicate samples, five replicates per 
sampling unit should usually be considered a minimum when bulk density 
is determined with cylinders (100 cm 3 or larger) or clods. Whether the 
sampling is carried out in a random pattern or on a predefined sampling 
grid, it is important for the interpretation of the measurements and their 
variability to note the visible structure of the individual samples and their 
localization with respect to different plant species and biological structures 
in the soil. In situations characterized by pronounced spatial patterns, such 
as many agroforestry systems, it may be useful to analyse the spatial 
variability of soil porosity with geostatistical methods (see Section 3.6), in 
which case a large number of measurements (at least 50) have to be taken 
together with their coordinates in the soil profile and/or on the soil surface. 
Between the collection of the undisturbed soil samples in the field and the 
porosity measurements, the samples are stored in the field-moist state at 
4°C to limit biological activity. 

In a more detailed approach to the analysis of soil structure, the 
variation of porosity as a function of the sample size within each sampling 
unit can be studied. This analysis usually reveals that the number of 
organizational levels within a soil sample increases with the sample size. If 
the total porosity of a soil sample does not vary with sample size, this 

200 M. Crimaldi etal. 

indicates a continuous, massive structure. On the contrary, if the porosity 
varies with sample size, it is possible to evaluate the different components 
of the total porosity, each of which is related to an organizational level of 
the soil structure. At least two such organizational levels can usually be 
distinguished (Stengel, 1979; Fies and Bruand, 1990). The assemblage of 
elementary particles within aggregates determines the textural porosity, 
and the assemblage of the aggregates determines, together with pores and 
cavities of biological origin, the structural component of porosity. For this 
type of analysis, bulk density is measured on samples varying between a 
few mm 3 and several hundreds of cm 3 in size, using the techniques 
described in Section 10.2. 

Pore size distribution 

A common approach to the analysis of soil porosity consists of the 
determination of pore size distribution, using either mercury injection or 
water desorption curves on undisturbed samples (Fig. 10.1). Both methods 
are based on Jurin's law (Eq. 10.2). If the pore space is approximated by 
a number of capillaries of variable size, but of simple form (cylindrical tubes 
or fissures with smooth and parallel planes), the capillary pressure P c , in 
Pa, of the water is related to the maximum equivalent size of the water- 
filled pores for each equilibrium state defined by the matric potential of 
the soil water: 

4v cos a 
Pc=Po-P„=— = ~h (10.2) 

where: P = atmospheric pressure; P w = water pressure (both in Pa); v = 
surface tension of water (75 x 10 -3 Nm _1 ) ; a = contact angle of water with 
the pore walls (0° if the water perfectly moistens the solid particles); d = 
equivalent size of the pore, corresponding to the diameter of a cylindrical 
capillary (therefore also 'equivalent diameter') or twice the distance 
between the two walls of a fissure (in m); and h = matric potential of the 
water (in Pa). 

The measurement of pore size distribution allows the total porosity of 
a soil to be subdivided into size classes of pores which differ in their 
ecological functions in the soil. Several classifications have been proposed, 
resulting in considerable confusion in terminology. Some authors divide 
the pore space into macroporosity (non-capillary pores) and microporosity 
(capillary pores) according to arbitrarily defined threshold values, such as 
-60 hPa. These should be adapted to each type of soil and the purpose of 
the study being undertaken (for the importance of macropores for water 
infiltration see Sections 7.1 and 11.4). A more comprehensive subdivision 
of pore space into fissures, transmission pores, storage pores and residual 

Soil Structun 


0.4 - 




\ 0.3- 


E ° 2 1 



p 0.1 - 


i 1 1 ■■■■■! i i 

Cumulative curves 

0.01 0.1 1 10 100 

0.0 ') frui ii| ' i i miii | i i hii i h i nnm | I | 

0.01 0.1 1 10 100 

Equivalent pore diameter (c/ urn) 



^ 0.3 H 

< 0.2 




Pore size distributions 



0.1 - 


Soil 2 

"""I i ""'"I ' """I ' '"""I i """'I ' ' u.u ' i n | i n» i i i | 

0.01 0.1 1 10 100 0.01 0.1 

Equivalent pore diameter (d eq , urn) 

mm] — r 


lllllll I I 


Fig. 10.1 . Comparison of pore-size distributions of two Amazonian Oxisols with 
clayey texture (soil 1, 91% clay) and sandy texture (soil 2, 9% clay) as determined 
by water desorption (solid lines) and mercury injection (dotted lines). V , pore 
volume; d. equivalent pore diameter 

202 M. Crimaldi etal. 

pores with limits at equivalent sizes of 500, 50, 5 and 0.5 (Jin has been 
proposed by Greenland (1979). An ecologically important division is at 
0.2 (im equivalent pore size, corresponding to the conventionally defined 
permanent wilting point (pF 4.2). Water in smaller pores, within the clay 
domains, is not available to many plants. The pore space <0.2 (im also 
corresponds well with the residual water in van Genuchten type models 
of soil hydraulic properties (see Section 1 1.4). For a further discussion of 
pore size classes and their functional properties, see Payne (1988). 

Water desorption method 

For establishing the water desorption curve, the soil samples are saturated 
with water on a suction plate or an equivalent device, and are then exposed 
to progressive desiccation. At each equilibrium state, the mass of the moist 
sample is measured to calculate its water content. The water volume 
obtained from the water content equals the volume of the water-filled 
pores whose maximum equivalent size is given by Jurin's law (Eq. 10.2). 
Different laboratory techniques are used to vary the matrix potential of 
the water within the large range between saturation and -10 6 hPa (pF 
6) (Tessier and Berrier, 1979; Bruand, 1990). These include suction plates 
(h > -10 2 hPa), pressure membranes (-10 1 hPa > h > -1.6 10 4 hPa), and 
desiccators wherein the relative humidity is controlled by a concentrated 
salt solution (h < -10 5 hPa). 

Mercury injection method 

The mercury injection technique is an alternative to the water desorption 
method for measuring pore size distribution (Lawrence, 1977). Mercury 
is a liquid which does not moisten the soil. It penetrates a porous, 
dehydrated and degassed sample if a pressure (P H ) that is inversely 
proportional to the size of the penetrated pores is applied: 

PHg = _4 E cose (1Q3) 

a e q 

where (1 = surface tension of mercury (0.480 Nut 1 ); = contact angle of 
mercury with the soil particles (141°), <i eq as in Eq. 10.2. 

Mercury injection is thus physically equivalent to the desorption of 
water; the mercury permeates the soil in the same way as air permeates a 
drying soil sample. The analysis of pore size distribution with mercury 
injection is carried out on small sample volumes (= 1 cm 3 ). The method is 
appropriate for pores of an equivalent size between 200 (im and 7.5 nm 
(for a maximum mercury pressure of 10 6 hPa). The technique is rapid and 

Soil Structure 203 

precise because the injection of the mercury into the sample is continuously 
recorded, contrary to the desorption of water, which is commonly 
measured only at distinct suction levels. However, the dehydration of the 
sample before the measurement provokes a certain shrinking of the soil 
and thus a modification of the pore space. This shrinking depends on the 
clay minerals and on the initial structure of the soil. The importance of 
this modification of the pore space can be evaluated by comparing the 
results from mercury injection with those from water desorption. In soils 
with kaolinitic clay mineralogy, the pore size distribution is not seriously 
altered by the desiccation, but there is nevertheless a slight effect on pore 
size classes (Fig. 10.1). This effect is not a pure artefact, because the 
shrinking of the soil is a physical soil characteristic that is interesting to 
interpret in relation to its structural organization. It should be noted that 
artificial drying methods such as lyophilization and the critical point 
method also affect the structure of the soil samples (Murray and Quirk, 
1980). It is thus preferable to air-dry and then oven-dry the samples and 
take the effect of shrinking into consideration in the interpretation of the 

Pore size analysis with mercury injection can reveal rather small 
modifications in the structural organization of the soil, as it is sensitive to 
changes in a wide range of pore sizes, including the pores between clay 
particles (residual pores), those between larger particles and clay 
microaggregates (storage pores) and a significant proportion of the 
structural pores, such as fissures, channels and cavities of biological origin 
(Fig. 10.2). However, because of the necessary drying of the soil before the 
measurement, the results can only give indications for a degradation or 
regeneration of soil structure, but cannot provide a reliable, absolute 
measure of the evolution of soil porosity. The plant-available water volume 
which is retained by pores of an equivalent size >0.2 |J.m is generally 
underestimated (Table 10.1). 

A problem that the water desorption and the mercury injection 
techniques have in common is that of bottlenecks in the soil pore space. If 
pores are accessible through narrower bottlenecks, it is the size of these 
which controls the entry of mercury or the desorption of water (Hillel, 
1998). The dimensions of the pores calculated from Eqs 10.2 and 10.3 can 
thus be underestimated. Comparing mercury injection with image analysis 
of a microaggregated soil, Colleuille (1993) found that a tenfold 
underestimation of real pore sizes with porosimetry can occur due to such 
bottleneck effects. 


M. Crimaldi ef al. 

0.10 - 

0.05 - 






Primary forest 

(depth: 0-5 cm) 

E 0.15 




°- 0.10 - 





(depth: 10-15 cm) 


Equivalent pore diameter (c/ eq , urn) 

Fig. 10.2. Examples for land-use effects on pore size distribution in clayey 
Amazonian Oxisols as revealed by the mercury injection technique: (a) effect of 
cattle trampling in a pasture; (b) effect of two leguminous cover crops (Pueraria 
phaseoloides and Desmodium ovalifolium) in an oil palm (Elaeis guineensis) 
plantation on a soil that had previously been compacted by mechanized 
deforestation (after Grimaldi et al., 1993). 

10.5 Measuring the Role of Soil Organic Matter in Aggregate 
Stability (C. Schroth) 

As discussed in Section 10. 1 and Chapter 4, trees can influence the stability 
of the aggregate structure of soils by increasing their organic matter 

Table 10.1. Comparison of macropore and micropore volumes (cm 3 g _1 ) as calculated from water desorption and mercury injection 





of micropores 


(200 |im 



> 0.2 urn) 


|j.m > 




Origin of the samples 






Primary forest 






Three-year-old pasture 

(manual deforestation) 






Pasture abandoned for 5 

















Mechanical compression 

(10 3 kPa) in the laboratory 

Mechanical deforestation 

and Desmodium ovalifolium cover crop 

Mechanical deforestation 

and Pueraria phaseoloides cover crop 















206 M. Crimaldi etal. 

content. This role of organic matter is particularly important in sandy soils 
where clay minerals and iron and aluminium oxides contribute little to the 
stabilization of the soil structure. Such soils are widespread in West African 
savannas. Based on the analysis of 495 soil samples from this region, a 
critical level of soil organic matter, S, was developed, from which the 
structural stability of the soil and its sensitivity to erosion can be deduced 
(Pieri, 1989): 

S = soil organic matter (%) / (clay + silt) (%) x 100 (10.4) 

A value of S<5 indicates structurally degraded horizons with great 
sensitivity to erosion; 5<S<7 indicates soils with great risk of structural 
degradation; 7<S<9 indicates soils with small risk of structural 
degradation; and indices >9 indicate soils with sufficient amounts of 
organic matter to maintain a stable structure. 

Of particular importance for the stabilization of soil structure are 
organic matter constituents with cementing properties such as polysac- 
charides, which are present in soil as strands or nets of polymers connecting 
clay particles. Soils rich in oxides and hydroxides of iron and aluminium 
can have a very stable aggregate structure, in part because strong bonds 
are formed between polymers and these constituents (Payne, 1988). These 
polysaccharides can be of plant, microbial or faunal origin. From the five 
monosaccharides that usually dominate in soils - glucose, galactose, 
mannose, arabinose and xylose - the second and third are believed to be 
mainly produced by microbes, and the latter two mainly by plants 
(Gregorich et al., 1994). Polysaccharides have been found to be more useful 
for explaining the stability of aggregate structure in arable soils than in 
grassland and forest soils, possibly because other humus fractions, roots 
and mycorrhizal fungi are more important for structural stabilization under 
perennial than under annual vegetation (Payne, 1988; Carter et al., 1994). 

Total polysaccharides in soils can be determined by hydrolysis with 
concentrated sulphuric acid, followed by dilute sulphuric acid. Omitting 
the first step recovers labile polysaccharides, which are most of the 
polysaccharides other than cellulose. Detailed procedures are given by 
Lowe (1993). An increase of dilute acid-extractable carbohydrates was 
correlated with increased aggregate stability under cover crops in an 
orchard soil (Roberson et al., 1991). In two Oxisols from the Brazilian 
cerrado, both cellulosic and non-cellulosic polysaccharides were related to 
macroaggregation (Neufeldt et al., 1999). In other studies, a hot-water 
(80°C)-extractable carbohydrate fraction was found to be a more sensitive 
indicator than acid-hydrolysable carbohydrates for changes in soil organic 
matter quality and aggregation as affected by cropping systems (Haynes 
and Swift, 1990). The relative merits of dilute acid- and hot-water- 
extractable polysaccharides as indicators of soil organic matter quality in 
relation to aggregation are still under discussion (Gregorich et al., 1994). 

Soil Structure 207 

For ferralitic soils in Queensland, Australia, an active carbon fraction, 
oxidizable with 33 mM potassium permanganate, was found to be a more 
sensitive predictor than total carbon for aggregate stability under rain (Bell 
et al., 1998). Relationships between this active carbon fraction and rainfall 
infiltration were used to determine critical levels of active carbon to prevent 
runoff for a given rainfall regime (Bell et al., 1999). 

10.6 Soil Micromorphology and Image Analysis 

(M. Grimaldi) 

The study of soil structure is greatly facilitated by the acquisition of images 
of different magnification with a stereo, optical or electron microscope. A 
specific terminology is used to describe the arrangement of mineral and 
organic soil constituents, the morphology of pores, and the origin of 
aggregates and pores (Brewer, 1964; Stoops and Jongerius, 1975; Bullock 
et, al., 1985; Fitzpatrick, 1990; Ringrose-Voase, 1991). These essentially 
qualitative descriptions are profitably combined with quantitative image 
analysis (Hallaire and Curmi, 1994; Ringrose-Voase, 1996), which is based 
on the concepts of mathematical morphology (Serra, 1982) and allows a 
full statistical evaluation of soil micromorphological data. 

Soil samples with undisturbed structure are collected from profile walls 
at depths which are chosen in accordance with the horizons and sub- 
horizons of the pedological profile. The samples are either air-dried, which 
provokes a certain transformation of their structure, or are kept in the 
field -moist state at 4°C. In the latter case, they are dehydrated by exchange 
of water by acetone, followed by impregnation with a polyester resin, to 
which a fluorescent pigment is added to visualize the pores under ultra- 
violet light (Hallaire, 1994). After the polymerization of the resin, the 
blocks are cut and polished to be studied under a stereomicroscope. Thin 
sections are prepared for analysis with optical or scanning electron 
microscopy. The use of backscattered electron scanning microscopy results 
in a more complete characterization of the pore space than that obtained 
by optical microscopy alone, as it also makes the pores between skeletal 
grains and clay domains visible, which, in fine-textured soils, are not 
accessible with other microscopic techniques. Only the very small pores 
within the clay domains remain invisible even with this method (Fies and 
Bruand, 1990). The method was used by Bruand et al. (1996) to study the 
changes of pore organization in the neighbourhood of roots. When bio- 
logical structures arising from the activity of different soil invertebrates or 
growing roots can be recognized (Eschenbrenner, 1986), the methods of 
image analysis are particularly useful for a quantitative characterization of 
the resulting spatial and temporal changes in the soil structure (Deleporte 
etal., 1997; Peres etal., 1998; Jegou et al., 1999; Barros et al., 2001). 

208 M. Crimaldi etal. 

Image analysis allows us to distinguish and quantify zones of higher 
and lower density. A large number of parameters related to the image as 
a whole (image parameters) or to an individual object (object parameters) 
can be used for the analysis (Ringrose-Voase, 1991). With respect to the 
pore space, the surface porosity, which is the intersection of the three- 
dimensional pore space with the image plane, is the image parameter that 
is the easiest to quantify. Each poroid, or apparently individual pore in 
the image plane under a given magnification, is characterized by form and 
size parameters. This allows calculation of the distribution of the surface 
porosity according to different size and form classes. A frequently used 
form index is given in the following equation: 

I = P 2 /(4kA) (10.5) 

where P is the circumference of the poroid and A its area. This allows 
rounded (/ < 5), elongated (5 < / < 25) and complex pores (/ > 25) to be 
distinguished. To these pore classes, different origins and, to some extent, 
functions can be ascribed. Elongated pores are mainly fissures that are 
produced by wetting-drying cycles or by mechanical stress from root 
growth. Very irregular pores are generally pores formed by the 
assemblage of aggregates and are often of biological origin. Rounded pores 
are usually galleries made by invertebrates (e.g. earthworms) or root 

A limitation of the use of micromorphological techniques for the 
analysis of soil structure is that the connectivity of the pore space, which 
is of fundamental importance for water and gas transport in the soil and 
thus for drainage and soil aeration, cannot be evaluated reliably due to 
the two-dimensional nature of the sections. For further information and 
methodological details see also Fox et al. (1993). 

Chapter 1 1 
Soil Water 

W.G. Teixeira, 1 F.L. Sinclair, 2 B. Huwe 3 and G. Schroth 4 

' Empresa Brasileira de Pesquisa Agropecuaria-Amazonia Ocidental, 
CP 319, 6901 1-970 Manaus, AM, Brazil; 2 School of Agricultural and 
Forest Sciences, University of Wales, Bangor, Gwynedd LL57 2UW, 
UK; 3 Institute of Soil Science and Soil Geography, University of 
Bayreuth, 95440 Bayreuth, Germany; 4 Biological Dynamics of Forest 
Fragments Project, National Institute for Research in the Amazon 
(INPA), CP478, 69011-970 Manaus, AM, Brazil 

11.1 Synopsis (F.L. Sinclair) 

Water is essential for plant growth and, as an inevitable consequence of 
opening their stomata to enable gaseous exchange during photosynthesis, 
plants transpire water, which they usually have to take up from the soil. 
The soil water content also has a pronounced influence on nutrient uptake 
from the soil as it affects root growth and the transport of nutrients to the 
root. Furthermore, it influences the availability of oxygen, microbial and 
faunal activity, leaching of nutrients and agrochemicals into the subsoil, 
and swelling and shrinking of certain clay soils. 

The water content of soil represents a balance between processes that 
add water to the soil, such as infiltration of rainfall, and processes through 
which water is lost from the soil, such as plant water use (transpiration), 
evaporation, runoff and drainage. Typically less than a third of the rainfall 
input is transpired in crop production systems in dry areas, indicating that 
there is scope for addition of trees to improve water-use efficiency. This 
may occur either directly, by capturing more water for plant production, 
or indirectly, by modification of the microclimate in ways that increase the 
transpiration efficiency of crops. This book is about soil fertility and so the 
methods discussed in this chapter focus on aspects of the water balance 
associated with the soil rather than with plants. In the synopsis, global 
water supply and demand are discussed, revealing water scarcity as an 
increasingly widespread agricultural constraint. The components of the 
water balance are then discussed with particular reference to how the 
incorporation of trees in agricultural practices may affect the overall 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 209 

F.L. Sinclair) 

210 W.C. Teixera ef al. 

efficiency of the use of rainfall and impact upon crop productivity. The 
next two sections explain how soil water content and water potential can 
be measured. Soil water content refers to how much water there is in soil, 
while its potential relates to the ease with which it can be extracted. The 
fourth section discusses measurements of soil hydraulic properties, such 
as the rate of water infiltration. The final section describes new techniques 
that use stable isotopes to estimate how much transpired water is being 
taken up from subsoil, as opposed to surface soil. 

Global trends in water resources and tree-crop interactions 

Although there is a vast quantity of water on Earth, most of it is saline, 
frozen or in underground aquifers, so that less than 1% is reasonably 
accessible fresh water (Wallace and Batchelor, 1997). This accessible fresh 
water is unevenly distributed and about a third of the world's land surface 
area is classified as arid or semiarid, indicating that water is a major 
constraint on agricultural production in these regions (UNEP, 1992). At 
present, about 7% of the world's population lives in areas where water is 
scarce, but it is anticipated that this proportion will rise tenfold over the 
next 50 years, largely because of the increasing water required to grow 
food for a burgeoning world population, which is growing particularly fast 
in drier regions (Wallace, 2000). Since about 75% of human use of fresh 
water is consumed in agriculture, the efficiency with which rainfall is used 
to produce food, especially where water is scarce, is of fundamental 
importance to sustaining a global balance between food supply and 

The efficiency of use of rainfall in crop production in semiarid areas 
is low, typically less than 30% for rain-fed cropping and less than 20% for 
irrigated agriculture (Wallace, 2000). This means that, in principle, 
integration of trees into agricultural fields could increase overall rainfall- 
use efficiency by more of the rain being used in plant transpiration because 
the trees use water not captured by crops, or because they lead to an 
increase in the amount of water infiltrating into the soil. In addition, trees 
could modify microclimate, leading to higher transpiration efficiency of 
crops, resulting in production of more dry matter per unit of water they 
transpire. But trees may also use a lot of water themselves and although 
they may access some water not available to crops, either from below the 
crop rooting zone or at times in the season when crops are not present, 
they may also compete with crops for water in surface soil. Similarly to 
tree-crop interactions for nutrients (see Chapter 5), those for water can 
be described in terms of facilitation (where trees improve crop water 
use through increased infiltration or transpiration efficiency), 
complementarity (where trees use water that would not have been used 

Soil Water 211 

by the crops) and competition (where trees use water that would otherwise 
have been used by the crops). As with nutrients, these forms of interaction 
often occur simultaneously. Research is urgently required to understand 
the circumstances, in terms of species combinations, the spatial arrange- 
ment of tree and crop components, site conditions and management 
practices, that result in facilitative interactions and complementary water 
use among trees and crops without incurring a lot of competition. For 
example, trees in boundary plantings may compete less with crops for soil 
water than evenly spaced trees in crop fields while retaining much of their 
beneficial effects on microclimate and infiltration. Some trees are able to 
grow in swampy and seasonally flooded areas, which are unsuitable for 
crops, in which case increased landscape water use may be achieved 
without any negative side effects on the availability of water to crops. 
Furthermore, there are species differences among trees in their rooting 
habit and their water use per unit of both root length and leaf area, and 
in how these respond to management interventions such as pruning (Jones 
et al., 1998). This makes choice of tree species and their subsequent 
management vital for achieving benefits of trees in terms of making 
efficient use of soil water without incurring too much competition for water 
with associated crops. 

A major research requirement is to measure impacts of different tree 
species and management regimes on soil water depletion and crop yield 
across a range of agricultural systems and site types. It is especially 
important to get this information for productive tree species that farmers 
may be interested in husbanding because they can derive income from 
them or because the trees contribute to meeting their subsistence needs. 

Impacts of trees on water balance 

When trees are present in crop fields they have a profound influence on 
most components of the water balance (Fig. 11.1), which means that soil 
moisture (9) will vary spatially in relation to distance from trees because 
of different inputs and uptake of water caused by their presence. Rainfall 
is normally the main water input, although in some situations run-on water 
from upslope areas can be important. Runoff irrigation systems in areas 
with insufficient rainfall for agriculture are specifically designed to 
maximize this latter water input (Lovenstein et al, 1991; Lehmann et al, 
1999a). Trees affect the amount and spatial distribution of rainfall (P ) that 
reaches the soil, with most of the area beneath tree crowns receiving less 
rainfall than open areas (P c ) because the tree crowns intercept some of the 
rainfall, which evaporates directly back to the atmosphere (7 t ), whereas the 
rest reaches the soil either as throughfall (P t ) or stemflow (P s ). Wallace 
(1996) estimated that for sparse tree canopies typical for tropical 


W.C. Teixera ef al. 

Fig. 11.1. Components of a water balance in a tree-crop system (see text for 
explanation of symbols). Subscripts t refer to crop and tree components, 
respectively (adapted from Wallace, 1996). 

agroforestry systems, interception losses are likely to be in the range of 
3-10% of total rainfall, with decreasing relative losses as total rainfall 
increases. Schroth et al. (1999b) measured 6.4% interception in 
agroforestry plots with different tree crops in Amazonia. Calder (2001) 
showed that the larger drop sizes in tropical convective storms reduce 
interception losses compared with trees in temperate climates. The division 
of rainwater into throughfall and stemflow during crown passage 
influences the spatial pattern of water inputs into the soil, and 
consequently soil water content and drainage. Certain tropical trees such 
as palms may concentrate considerable amounts of rainwater as stemflow 
in their proximity, thereby possibly reducing its availability to associated 
crops. The water input into a 2 m 2 area around the stems of peach palms 
(Bactris in an Amazonian agroforestry system was 152% of open- 
area rainfall (Schroth et al., 1999b). 

Soil Water 213 

Another major effect of the tree crowns is to intercept radiation and 
so reduce the energy reaching the soil or crop beneath. This can 
substantially reduce evaporation of water from bare soil under trees (E t ), 
which is important because more than 30% of total rainfall is typically lost 
through evaporation from soil in rain-fed cropping systems in the semiarid 
tropics (Wallace, 2000). The presence of 50% tree cover in an agroforestry 
system in semiarid Kenya was found to reduce cumulative soil evaporation 
over a season from about 55% of the rainfall input to 39% (Wallace and 
Batchelor, 1997; Wallace et a!,., 1999). Soil evaporation is also reduced by 
tree litter and mulch (Wallace, 1996). 

Trees may also enhance infiltration (F t ) over that occurring on areas 
with crops alone (F c ) by protecting the soil surface from rain splash with 
their litter and mulch and increasing soil faunal activity, thereby reducing 
runoff (R t ). Where trees are planted on the contours, runoff water from 
cropped surfaces may infiltrate under the trees, where infiltration rates 
may be higher than under the crops (Kiepe, 1995; see Chapter 17). 
Drainage may also be lower in the vicinity of trees (Z) t ) because water is 
taken up from more of the vertical extent of the soil profile than in areas 
where only crops are present and this water is subsequently transpired 
(T t ). In dry areas, trees may be able to access water from several tens of 
metres depth (Leakey et al. , 1999). Water uptake by trees may also reduce 
subsurface lateral flow of water (R s ). Lateral drainage occurs on slopes and 
may be especially important where sandy topsoils overlie clayey subsoil 
horizons of lower permeability, a situation commonly encountered in some 
tropical regions (Chauvel et al., 1987). Runoff and drainage typically 
account for 40-50% of rainfall input in semiarid rain-fed cropping systems 
(Wallace, 2000). Tree root systems may also redistribute water vertically 
in the soil profile by absorbing water from wet soil at one depth and 
exuding it in dry soil either higher or lower in the profile (H t ). This was 
first observed in the upward direction, where shrubs took up water from 
moist soil at depth and exuded it in dry soil near the surface, enabling 
nutrient uptake by shrubs and growth of associated herbaceous plants, 
and was termed hydraulic lift (Caldwell and Richards, 1989). More recently 
the process has been found to operate in reverse with water travelling 
down roots from wet surface soil to be exuded at depth, facilitating root 
growth through dry soil to reach water and nutrients that are available at 
depth in some arid ecosystems (M. Smith et al., 1997; Schulze et al., 1998). 

Over an annual cycle the trees will also take up and transpire water at 
times when the crop is not present or when it has only a small leaf area 
and root length. The extent of the seasonal complementarity of water use 
will depend on the rainfall distribution at the site and the phenology of 
leaf development and retention by the trees. The reverse phenology of 
Faidherbia albida, a tree species of African savannas which loses its leaves 
during the wet season, is likely to lead to temporal complementarity of 

214 W.C. Teixera ef al. 

water use with annual crops. Roupsard et al. (1999), monitoring water use 
of adult F. albida trees in a Sudanese parkland, found that the tree roots 
accessed a water table at about 7 m depth and so were able to be productive 
during the dry season, with the fraction of annual rainfall used by trees 
remaining below 5%, indicating little competition with annual crops grown 
in the wet season. Similar effects can be achieved through strategic crown 
pruning of trees during dry spells or phases with high crop water 
requirements (Smith et al, 1998). Surprisingly little is known about the 
phenology of many trees with agroforestry potential, so measuring 
seasonal water extraction from soil by trees with different phenology, to 
find species that will exhibit complementarity with different cropping 
patterns under various rainfall regimes, is a key area for research. 

Transpiration efficiency 

In addition to the impacts on the amount of water captured for 
transpiration, trees may also increase the transpiration efficiency of crops 
by reducing the saturation deficit of the air beneath their crowns. 
Overhead tree shade generally reduces radiation receipt and air 
temperature, and increases humidity experienced by the understorey. The 
presence of tree crowns may also reduce wind velocity (Green et al., 1995) 
and so reduce mixing of air, maintaining a low saturation deficit over the 
crop. This means that less water will be transpired per unit of dry matter 
produced by the crop because, when stomata open to allow gaseous 
exchange in photosynthesis, less water evaporates from leaves (Wallace 
and Verhoef, 2000). Effects of shelterbelts may be more complex because 
the shelter is provided laterally so that wind speed is reduced but there is 
no overhead shade to reduce radiation receipt. This can lead to higher 
temperatures and saturation deficits at leaf surfaces in the lee of the 
windbreak and hence greater water loss per unit of dry matter 
accumulation (Brenner et al, 1995). Shading could also improve 
transpiration efficiency by reducing heat stress that unshaded crops 
frequently experience in areas of high incident radiation and temperature 
(Ong etai, 2000). 

The measurement of the plant and atmospheric components of the 
water balance in tree-crop systems has been discussed and described by 
Wallace (1996). In recent years major advances have been made in direct 
measurement of water uptake by plants using heat as a tracer to measure 
sap flow rate. The various techniques for measuring sap flow in stems of 
trees and crops have been reviewed by Smith and Allen (1996) and their 
application to tree and crop roots by Fernandez et al. (2000). Another 
recent development of particular relevance to agroforestry, the use of 
stable isotopes to estimate the proportion of plant water uptake from 

Soil Water 215 

subsoil as opposed to surface soil, is discussed in Section 11.5. 
Measurement methods for soil water content, water potential and soil 
hydraulic properties are discussed in the following sections. 

11.2 Methods for Soil Water Content (W.G. Teixeira, 
B. Huwe) 

Measurements of soil moisture express the amount of water in a given 
mass or volume of soil. The water content is usually reported in cm cnr 3 
or in g g- 1 relative to the dry soil. Soil water content (and water potential) 
are highly variable in both space and time, changing markedly after each 
rainfall event. Addition of trees to cropping systems exacerbates this 
variability. Ideally, therefore, it is desirable to record soil water data 
automatically at short time intervals for a large number of points in 
horizontal and vertical positions in the soil. In large experiments with 
numerous measurement points the cost of equipment to make automatic 
measurements over the whole experiment may be prohibitively high. A 
useful compromise between minimizing cost and maximizing temporal 
and spatial resolution is to conduct infrequent, manual readings at many 
measurement points within a study site combined with frequent, 
automated readings at a few points. 

Gravimetric method 

The gravimetric measurement of soil water content involves the weighing 
of collected soil samples before and after drying at 105°C to constant 
weight. The principle of the method is simple but the sampling and 
laboratory procedures are labour intensive. The method is destructive, 
which makes it impossible to resample the same site and automate the 
acquisition of data. The measurement can be carried out on either 
disturbed soil collected with a soil auger or undisturbed soil collected with 
steel cylinders of known volume. The deviations normally found among 
data from disturbed and undisturbed samples are related to the fact that 
the mass-based values from disturbed samples are recalculated to 
volumetric values by means of the measured bulk density, thus introducing 
a new source of error. For gravimetric methods, normally taken as the 
'correct' value of soil moisture, the principal sources of inaccuracy are 
(Gardner, 1986): 

• an inappropriate sampling scheme; 

• an inadequate number and volume of samples; 

• uncertainties in the equilibrium time when drying the sample (if 
repeated weighing of the sample is not conducted); 

216 W.C. Teixera ef al. 

• the presence of colloidal material that retains water even when exposed 
to high temperatures; 

• the presence of organic material that can oxidize or volatilize, leading 
to specific problems in some organic soils; 

• difficulties in maintaining a constant temperature in the oven; and 

• the precision of the balance used to weigh samples. 

Soil samples can also be dried by microwaves but this method may not 
give reliable results for soils containing significant amounts of mica, 
gypsum, halloysite, montmorillonite or other hydrated materials common 
in organic soils (Liu and Evett, 1990). 

The direct gravimetric method is the most commonly used method in 
soil water studies in tropical agroforestry. The effect of hedgerow 
intercropping on soil moisture determined by gravimetric methods has 
been investigated in numerous studies in the humid, subhumid (Lawson 
and Kang, 1990) and semiarid tropics (Lai, 1989; Rao et al, 1991), focusing 
on the balance between beneficial and competitive effects between trees 
and crops. For example, detailed measurements of the soil water profile 
at Machakos, Kenya, substantiated the importance of competition of 
hedgerows for water in semiarid environments, which often leads to 
reduced crop yields despite beneficial windbreak effects of the trees (Lai, 
1989; Rao et al., 1998). Sequential coring and gravimetric determination 
of soil water content were used by Jones et al. (1998) to show differences 
among tree species in water uptake from surface soil, differences in how 
species responded to shoot pruning and hence their competitiveness with 
crops and the extent to which this could be controlled by management 
interventions. Biweekly gravimetric determination of soil moisture was 
also used in a nutrient leaching model for an agroforestry system with 
cocoa and shade trees (either Cordia alliodora or Erythrina poeppigiana) in 
Costa Rica (Imbach et al., 1989). 

Neutron scattering 

The neutron-scattering method is a non-destructive procedure developed 
in the 1950s and still in use in many research centres. It uses the principle 
of neutron thermalization. High -energy neutrons that are emitted from a 
radioactive substance, such as 241 americium-beryllium, slow down and 
scatter by elastic collisions with hydrogen nuclei. The rebounding neutrons 
are measured with a detector (Greacen, 1981; Gardner, 1986). The sphere 
of influence of the emitted neutrons is roughly spherical with a radius of 
about 10 cm in wet soil and 25 cm in dry soil (Hillel, 1998). When 
measurements are carried out close to the soil surface, the sphere may 
extend into the air, making measurements in the top soil both unreliable 

Soil Water 217 

and a safety hazard. The measurement device is inserted into the soil 
through thin-walled aluminium access tubes, which are installed in holes 
in the soil created with a soil auger. Because of interactions with other 
atoms in the soil, such as cadmium, lithium, boron and chlorine, and the 
fact that hydrogen is present in soil organic matter and as structural water 
in some soil mineral components, extensive calibration against gravimetric 
water contents maybe required (Greacen, 1981; Gardner, 1986; Grimaldi 
et al., 1994). The calibration equation is normally linear, but to determine 
the slope accurately it is essential to sample a sufficiently wide range of dry 
and wet points (Greacen, 1981). Measurement bias resulting from 
instrument drift is partially avoided by the normalization of the observed 
count rate by the count rate in pure water, which is strongly recommended 
(Williams and Sinclair, 1981). In agroforestry research, changes in soil 
organic matter content and thus hydrogen concentration of the soil 
provided by sources other than water may necessitate periodic 
recalib ration of the method. 

An advantage of neutron probes is independence of the measurements 
from temperature and pressure (Kutilek and Nielsen, 1994). The major 
disadvantages are radiation hazard and the infeasibility of making 
measurements near the soil surface. The radioactive device must be 
handled with extreme caution and stored properly and cannot be installed 
in the field for automated data collection. Safety regulations require 
training of the users and government licences for the ownership, transport 
and use of a radioactive source. 

Examples for the use of the neutron-scattering method in agroforestry 
research include Singh et al. (1989) and Govindarajan et al. (1996) in 
studies of tree-crop interactions in hedgerow intercropping in India and 
Kenya, respectively, and Eastham et al. (1994) in studies of water-use 
efficiency of fodder trees in silvopastoral systems in Australia. 

Time domain reflectometry (TDR) 

This method is increasingly used as a non-invasive technique for the 
determination of the volumetric water content of soils. It is based on the 
determination of the dielectric constant (e) of the soil through the 
measurement of the propagation velocity of electromagnetic waves, using 
the large difference of e between water (=81), air (=1), mineral 
constituents of soil (3-5), and frozen or bound water (= 3.2). The dielectric 
constant is measured with probes that are installed in the soil, down to a 
depth of several metres, if required, and that are either permanently 
connected to a datalogger or are temporarily connected to a mobile device. 
Certain devices allow instantaneous measurements of topsoil moisture with 
minimum soil disturbance with mobile probes that are inserted vertically 

218 W.C. Teixera ef al. 

into the soil when measurements are to be made. It is also possible to use 
the TDR technique with access tubes as in the neutron-scattering method. 

TDR offers the possibility of continuously monitoring changes in soil 
water content, even at remote sites. The technique involves a relatively 
large initial investment for the acquisition of the equipment, but it can be 
easily installed in measuring networks with automated data collection and, 
for large numbers of measurement points, it can be less expensive than 
neutron probes, although it is generally more expensive than frequency 
domain methods (see below). The role of the TDR technique in 
agroforestry research is likely to increase in the future. 

It was initially believed that a single universal equation could be found 
relating soil water content (9) to e. However, subsequent work showed that 
various factors may influence the measurement of e, so that site-specific 
calibrations may be necessary for improving the accuracy of 9 estimates, 
especially in clayey soils, soils with low bulk density (Roth et al., 1992; 
Dirksenand Dasberg, 1993; Malicki etal., 1996) and soils with high organic 
matter content (Herkelrath et al., 1991; Roth et al., 1992). The empirical 
calibration equations generated in temperate soils frequently give 
inadequate accuracy in tropical soils (Dirksen and Dasberg, 1993; Weitz et 
al., 1997). For example, in a central Amazonian Oxisol with about 79% 
clay content and a bulk density around 9.9 Mg rrr 3 , use of the built-in, 
temperate-zone calibration of a TDR device led to errors of 9.19 m 3 rrr 3 
(Teixeira, 2991). The temperature effect on determination of e is reported 
to be linear (Pepin et al. , 1995; Perssonand Berndtsson, 1998) and should 
be corrected where large temperature fluctuations occur, especially near 
the soil surface. Under saline conditions or in the presence of large 
amounts of fertilizer, which can interfere in e measurements (Dalton, 1992; 
Kim et al., 1998), more accurate measurements can be obtained with coated 
buriable waveguides. In some tropical soils, high iron content and the 
presence of magnetic minerals can likewise interfere in e determinations 
(Roth et al., 1992; Robinson et al., 1994). Calibration procedures using 
empirical (Topp et al., 1989; Malicki et al., 1996) or physically based (Roth 
et al., 1999; Weitz et al., 1997) relationships between e and volumetric soil 
moisture can be found in the literature. 

When installing the probes in the field, air pockets, stones and the 
channelling of water along the probes need to be avoided. In agroforestry 
research, care should also be taken to avoid bias due to small-scale 
variations in soil properties caused by vegetation or management actions 
that may affect the soil water content. Soil temperature and bulk density 
may vary over short distances in the field because of differences in leaf 
litter, root density or tillage, and their influence on the TDR calibration 
should be checked at the beginning of a measurement programme. 

Because of the recent development of the TDR technique, few 
agroforestry applications have so far been reported. Yunusa et al. (1995) 

Soil Water 219 

used the method together with neutron scattering to analyse plant water 
use and deep drainage in a silvopastoral experiment with pines in New 
Zealand. V.L. Grime and F.L. Sinclair (unpublished) used the technique 
together with gravimetric methods in a large agroforestry experiment with 
dispersed trees and sorghum on vertisolic soils on the edge of the Sahel 
in northern Nigeria, while Wallace (1996) has used TDR in semiarid 
conditions in Kenya and Teixeira (2001) in tree-crop agroforestry in 
central Amazonia. 

Frequency domain reflectometry (FDR) or capacitance method 

This method is based on the measurement of the resonance frequency 
when high-frequency electrical waves are created around a sensor (Dean 
et al., 1987). The sensor basically consists of a pair of electrodes that form 
a capacitor, with the soil acting as the dielectric medium. Alteration in soil 
water content is detected by changes in the operating frequency. Recent 
systems use probes permanently buried in the soil or mobile probes 
inserted through access tubes as in neutron-scattering measurements. 

Advantages of the capacitance methods are the possibility of 
connection to conventional data loggers and their relatively low cost. The 
method requires that probes are calibrated for each soil. Separate 
calibrations are necessary for positions or horizons differing in bulk 
density. Many of the advantages of TDR are also valid for FDR including 
rapid, repeatable measurements and near-surface measurements. The 
disadvantages are greater sensitivity to salinity and to air gaps or cracks in 
the soil around the electrodes (Evett and Steiner, 1995; Paltineanu and 
Starr, 1997). Careful installation of access tubes is, therefore, essential. 

Other methods 

These include gamma ray attenuation (Gardner, 1986) and remote sensing 
by ground-penetrating radar (Weiler et al., 1998). It is also possible to 
estimate soil moisture indirectly from soil water potential through 
parametric models (Campbell, 1974; van Genuchten, 1980). This approach 
has been used in a hedgerow intercropping experiment with Leucaena 
leucocephala and Flemingia macrophylla in Zambia (Chirwa et al., 1994b). It 
has to be used extremely carefully, as hysteresis can lead to incorrect 
estimates of soil water content (see Section 1 1.3). 

Table 11.1 summarizes the methods for evaluating soil moisture 

Table 11.1. Methods for evaluating soil moisture content. 





(m 3 rrr 3 ) 

(m 3 rrr 3 ) 

Collection of data 


Weighing of soil 
samples before and 
after oven drying 



Manual, destructive 

Neutron scattering 

Determination of 
rebounding thermalized 



Electrical, manual 

Time domain 

Determination of the 



Electrical, manual or 


velocity of propagation 
of electromagnetic 
waves in dielectric 

in mineral 

soils; ±0.07-0.18 

in organic soils 

automatic readings 

Frequency domain 

Determination of resonant 



Electrical, manual or 


frequency in dielectric 

automatic readings 


Accuracy of balance ±0.01 

Calibration is necessary; 
recalibration could be 
necessary with changes in soil 
organic matter 
Specific calibration is 
necessary, especially in organic 
and clayey soils; cable and probe 
rods should be periodically checked 

Sensitivity is concentrated near 
the probe; recalibration is 
recommended if the sensor is 
changed during the experiment 

Soil Water 221 

11.3 Methods for Soil Water Potential (W.G. Teixeira, 
B. Huwe) 

The soil water potential can be interpreted as the work (per unit mass, 
weight or volume) necessary to remove water held in the soil under the 
influence of matric (or capillary), osmotic and gravitational forces. It is 
commonly measured in centimetres of water column or its decadic 
logarithm, the pF value. Information on the soil water potential is essential 
in studies of water transport, including water uptake by plants. The 
relationship between the matric potential (or pressure head) and water 
content of a soil is a function of its pore size distribution (Fig. 11.2; see also 
Section 10.4). At a given water potential, clayey soils normally have a 
greater water content than sandy soils, but they also hold a larger part of 
the water in very fine pores, where the water is not available to plants. The 
relationship between soil water content and water potential is also 
influenced by soil structure (Fig. 11.2). The available water capacity of a 

Fig. 11.2. Typical soil water retention curves of a sandy soil, a clayey soil and a 
well-structured, clayey Oxisol. For the latter soil, note that the pattern of water 
release indicates a bimodal distribution of soil pores. The vertical lines indicate 
field capacity (pF 1 .8) and the permanent wilting point (pF 4.2). 


W.C. Teixera ef al. 

Matric suction (m) 

Fig. 11.3. Hysteresis in the relationship between water content and potential 
during the wetting and drying of a clay loam soil in the field (adapted from 
Marshall and Holmes, 1979). 

soil is defined as the difference between field capacity and permanent 
wilting point. The held capacity is the amount of water held by the soil 
against gravity, after excess water has drained away. The exact value is a 
matter of definition and is conventionally measured at a matric suction 
somewhere between 5 and 33 kPa. The permanent wilting point 
characterizes the water content at which all water is held in very fine pores 
from which it cannot be extracted by plant roots (see Section 10.4). It is 
conventionally measured at a matric suction of 1.5 MPa (pF 4.2), although 
the true value differs between plant species (Gregory, 1988). 

The relationship between water content and water potential of a soil 
displays hysteresis, which means that soil that has reached a certain water 
potential by wetting will have a lower water content than soil that has 
reached the same water potential by drying (Fig. 11.3). A detailed 
discussion of this phenomenon is given by Hillel (1998) and Iwata et al. 

There are numerous techniques to measure the soil water potential. 
Generally, these involve the measurement of some property of the water 
in a reference medium that is in thermodynamic equilibrium with the 
liquid or gas phase of the water in soil. 


These are the most common instruments for measuring the soil water 
potential in the field. They consist basically of a porous cup in the soil 
connected to a water-filled tube that transmits the soil water potential to 
a measurement device such as a manometer or electrical pressure 
transducer. Portable pressure transducers (tensiometers) can reduce costs 

Soil Water 223 

when a large number of instruments are needed. However, they are 
sensitive to temperature fluctuations and may require correction. The 
air-entry capacity of the commonly used porous cups restricts the 
measurement range of tensiometers to suctions less than about -85 kPa at 
sea level with an accuracy of approximately 0.1 kPa (Cassel and Klute, 
1986). At higher altitudes the reduced atmospheric pressure causes further 
reductions in the operating range. Because the absolute pressure inside 
the tensiometer includes the hydraulic head of the water column in the 
tube, the maximum depth to which a conventional tensiometer can be 
usefully operated below the soil surface is about 4 m (Livingston, 1993). 
At lower depths, tensiometers can be operated from soil pits. 

Tensiometers are often installed in groups at different depths. Before 
installation, each tensiometer should be calibrated as shown by Livingston 
(1993) and Cassel and Klute (1986). Intimate hydraulic contact between 
the ceramic cup and the surrounding soil is essential for reliable 
measurements. In most situations the tensiometer can be installed in a 
tight access hole. Sometimes installation at an angle may be helpful to 
prevent water flow along the tubes, but the corresponding reduction of 
the pressure head in the tube needs to be considered. In experiments of 
long duration with trees or perennial crops, preferential root growth 
around the ceramic cups can lead to unrepresentative readings requiring 
periodic reinstallation of the tensiometers. Frequent maintenance is 
required under hot and dry conditions, especially in sandy soils or 
shrinking clays. As soils dry out, air bubbles appear in the tensiometer, 
which do not obviate the measurement but reduce its sensitivity. They can 
be removed by applying suction or refilling the tensiometers with degassed 
water. Tensiometers are also susceptible to errors induced by temperature 
fluctuations and, therefore, the above-ground parts should be shielded 
from direct exposure to the sun. In hot regions it is recommended that 
readings are taken from tensiometers in the morning and always at the 
same hour. Where freezing temperatures occur, tensiometers have to be 
removed or emptied during the cold season. 

In the field, the variance of tensiometer readings often increases with 
the mean tension (Greminger et al., 1985; Saddiq et al, 1985; Hendrickx 
et al, 1990), and this may cause problems with some statistical tests and 
require transformation of data prior to analysis. The variance of 
tensiometer readings can be reduced by increasing the cup size (Hendrickx 
et al, 1994). 

The main problem with tensiometers is that they operate only in the 
relatively wet range of soil water potential, so their use in combination with 
other methods may be necessary. A recent capacitive technique allows 
measurements of soil water potential in the range 0-1000 kPa 
(Equitensiometer, Delta T, Cambridge), although at reduced precision. 
Other field methods for the dry range of soil water tension include heat 

224 W.C. Teixera ef al. 

dissipation (Phene et al., 1992; Reece, 1996) and soil psychrometry 
(Rawlins and Campbell, 1986). 

Tensiometry has been used in agroforestry research to determine 
temporal and spatial patterns of soil water potential and indirectly the 
activity of plant roots. Examples are the measurement of moisture 
depletion patterns under hedgerow intercropping with different tree 
species in Nigeria (Kang et al., 1985; Lawson and Kang, 1990), effects of 
boundary plantings on water availability in an adjacent crop field in Togo 
(Schroth et al, 1995a), interference between neighbouring plots through 
lateral tree roots in Nigeria (Hauser and Gichuru, 1994) and tree-crop 
interactions in runoff agroforestry in semiarid northern Kenya (in 
combination with gypsum blocks) (Lehmann et al, 1999a). 

Gypsum block sensors 

These sensors consist of two electrodes embedded in a porous block of 
gypsum (CaS0 4 ). When installed in the soil, the electrical resistance (Q) 
measured between the electrodes is determined by the water content of 
the blocks at equilibrium with the moisture potential of the surrounding 
soil. The soil water potential can be inferred with proper calibrations. 
Gypsum blocks are probably the cheapest sensors of soil water potential; 
they can be easily installed and connected to a data logger. 

For calibration, blocks are equilibrated with the respective soil at 
different water potentials, normally using a pressure plate. Individual 
calibrations are necessary for each block and for each soil horizon if these 
differ significantly. The calibration equations normally show non-linear 
inverse relationships between resistance and soil moisture potential 
(Campbell and Gee, 1986; Livingston, 1993). Temperature correction is 
necessary when the calibration temperature differs from that in the field; 
correction factors are normally supplied by the manufacturer. The rather 
high solubility of the gypsum used in the blocks leads to changes of the 
calibration parameters with time, especially in acid or alkaline soils and 
where the water table is frequently at high levels. Likewise, changes in the 
properties of the soil matrix may require periodic recalibration in long- 
term experiments. 

Although gypsum blocks can measure soil water potential in the range 
between -50 and -1500 kPa, this internal range may vary according to 
technical characteristics of fabrication. With calibration and temperature 
correction, the precision of soil moisture potential measurements is likely 
to be around ±0.1-0.5 MPa (Livingston, 1993). The low sensitivity of 
gypsum blocks in the dry range and restricted contact between the block 
and the surrounding soil make them unsuitable for use in sandy soils or 
shrinking clays (Gardner, 1986). The presence of salts such as fertilizer 

Soil Water 225 

leads to an overestimation of soil wetness (Young and Warkentin, 1975). 
Agroforestry applications of gypsum blocks include Huxley et al. 
(1994), who monitored the position of wetting and drying fronts in the soil 
profile in a tree-crop interface experiment with uncalibrated blocks in 
Machakos, Kenya, and Lehmann et al. (1999a) in northern Kenya. 

Granular matrix sensor 

The granular matrix sensor (Watermark Soil Moisture Sensor, Irrometer 
Co., Riverside) operates with the same electrical resistance principle as 
gypsum blocks. It also contains a wafer of gypsum embedded in a granular 
matrix to minimize salinity effects (Eldredge et al., 1993). The granular 
matrix sensors reduce some of the problems encountered with gypsum 
blocks. Of particular significance is their greater stability and thus longer 
lifetime in the soil. They exhibit sensitivity to soil water potential over a 
range from to -200 kPa and can, therefore, be used in drier soils than 
tensiometers and in wetter soils than gypsum blocks. This makes them 
adaptable to a wider range of soil textures than gypsum blocks. However, 
granular matrix sensors also require proper calibration and temperature 

Other methods of measuring soil water potential are described in 
Hillel (1998) and Klute (1986). 

11.4 Methods for Soil Hydraulic Properties (W.G. Teixeira, 
B. Huwe) 

The mobility of water in the soil is determined by soil hydraulic properties, 
particularly its saturated and unsaturated conductivity, which are functions 
of the soil water content or water potential (Fig. 11.4). These soil 
characteristics influence processes such as water infiltration (and thus 
runoff and erosion), its transport to plant roots, and leaching of nutrients 
and pesticides. They are thus an essential input in models of water and 
nutrient cycling. Hydraulic properties are a function of the number, 
volume, morphology, connectivity and stability of soil pores. They are 
affected by soil characteristics that influence the cohesive forces between 
soil constituents, such as pH, cation exchange capacity, iron and 
aluminium oxides, and organic matter (Hillel, 1998). Hydraulic properties 
are spatially and temporally variable and are easily modified by factors 
such as tillage, fertilization, use of mulch and cover crops, root growth, 
and faunal and microbial activity. 

In this section, some field methods for evaluating soil hydraulic 
properties above the water table are described. Laboratory techniques and 

226 W.C. Teixera ef a/. 

Fig. 11.4. Relationship between hydraulic conductivity (K) and matric suction 
evaluated with a tension infiltrometer from the soil surface near two tree crops on 
a central Amazonian Oxisol. 

methods below the water table are given by Stephens (1996), Carter (1993) 
and Klute (1986). One of the most important considerations in measuring 
hydraulic properties in the field is their large spatial variability. The 
presence of macropores or soil cracks commonly found in certain tropical 
soils further increases this variability. For comparisons between different 
experimental treatments, it is often better to have numerous measurement 
points with smaller precision than just a few highly accurate data points. 

Infiltrability and saturated hydraulic conductivity 

Infiltration rate refers to the vertical entry of freely available water into a 
soil surface. It should not be confused with hydraulic conductivity or 
permeability, which is a measure of the ability of a soil to transmit water 
in a three-dimensional system. In infiltration experiments, the infiltration 
rate tends to be numerically equal to the saturated hydraulic conductivity 
if the hydraulic gradient of the soil is unity. This condition is frequently 
approximated for homogeneous and isotropic soils (Black et al., 1969; 
Libardi et al, 1980). If incorrect, the assumption of a unit gradient can be 
a source of error in the calculation of the hydraulic conductivity (Ahuja et 
al, 1988; Reynolds and Zebchuk, 1996). Common causes of divergence 
from the unity gradient include entrapped air within the narrow pore 
space when an initially dry soil is wetted, and the occurrence of horizontal 
flows (Collis-George, 1977; Mbagwu, 1995). Nevertheless, for many 

Soil Water 227 

practical problems of large-scale significance, the mean infiltration path 
over a sufficiently large area is approximately one-dimensional in the 
vertical direction, especially in relatively homogeneous soils (Stephens, 

In tropical climates, the exposure of the soil surface frequently leads 
to the formation of a superficial soil crust. To verify if such a thin, 
superficial layer of low permeability controls the infiltration rate, 
measurements should be made first under natural surface conditions and 
then after the surface layer has been carefully removed. Worms tend to 
move upwards and crawl out of the soil when the surface is covered with 
water (Bouwer, 1986). The resulting open wormholes can greatly increase 
infiltration rates, particularly if the test is carried out over a long time. 
Prewetting the site before the measurements can alleviate this problem, 
which is frequently encountered in the tropics. The temperature and 
chemical composition of the water used for infiltration measurements 
should be the same as those of the soil water to avoid dissolution of soil air 
in the infiltrating water (Bouwer, 1986) and changes of the flocculation 
status of the clay particles (Hillel, 1998). If possible, rainwater from the 
site should be used to avoid such problems. The conditions of the 
measurements such as water quality, temperature, soil moisture and 
surface conditions, as well as the exact procedure of the prewetting and 
infiltration, should be recorded so that they can be duplicated in 
subsequent measurements. 

Double-ring infiltrometer 

The double-ring infiltrometer consists of two concentric metal cylinders 
that are pushed or driven a small distance into the soil. The cylinders are 
filled with water, and the infiltration rate is measured until it reaches a 
constant value in the inner ring (Bouwer, 1986). The simplicity and low 
cost of the method are its main advantages. 

As the head of water ponded on the soil surface affects the infiltration 
rate, a constant water level should be maintained in the cylinder with a 
float valve or a mariotte siphon arrangement. To prevent leakage between 
the cylinders, equal water levels must be maintained in both cylinders. A 
multiple automated system can be constructed to allow simultaneous 
measurements with several devices (Matula and Dierksen, 1989; 
Maheshwari, 1996). 

The use of a double ring, with measurements confined to the inner 
ring, minimizes errors due to non-vertical flow at the edge of the cylinder. 
However, in many tropical soils the flow below the infiltrometer is often 
not straight downward but diverges laterally. Hence the infiltration rate 
as measured in the infiltrometer will be greater than the true infiltration 
rate for vertical flow. To reduce this source of error, the outer cylinder 

228 W.C. Teixera ef al. 

should be as large as possible, even though big rings are cumbersome and 
require a large volume of water. To reduce the lateral divergence of the 
flux, the soil sites to be tested should be artificially prewetted, or the 
measurement should be carried out after rain has fallen. If the measure- 
ment cannot be carried out on the same day as wetting the soil, the site 
should be covered with a plastic sheet or plant residues to prevent surface 
drying. A restricting layer deep in the profile can cause overestimation of 
the infiltration rate, because lateral flow above this layer may be greater 
under measurement conditions than when the entire soil surface is 
inundated and all water has to move straight down through the soil and 
the restricting layer (Bouwer, 1986; Kutilek and Nielsen, 1994). In this 
situation it is preferable to use another technique such as the borehole 
permeameter (see below). 

In addition to lateral flow of infiltrating water, several other factors 
can affect the results from double-ring measurements. The soil compaction 
caused by the insertion of the cylinders can lead to a reduction of the true 
infiltration rate. On the other hand, the measured infiltration rate will be 
overestimated if a surface crust or other restricting layer at or near the 
surface is disrupted by the installation of the infiltrometer, or if there is 
imperfect contact between the restricting layer and the inside cylinder wall 
(Bouwer, 1986). 

The commonly used equations to describe infiltration results are the 
Philip's two-parameter equation, Green-Ampt equations and the empirical 
equations of Kostiakov and Horton. The applicability and utility of these 
different equations are discussed by Hillel (1998), Jury et al. (1991), Kutilek 
and Nielsen (1994) and Libardi (1995). Recently, methods to calculate 
infiltration from a single-ring infiltrometer using scaling approaches have 
been published (Wu and Pan, 1997; Wu et al, 1999). 

An extension of the double-ring method that allows testing of the 
assumptions of the infiltration measurement and subsequent deter- 
mination of unsaturated conductivity has been developed by Ahuja et al. 
(1976, 1980). The double-ring infiltrometer is used in combination with 
two multiple-depth tensiometers installed in the inner and outer rings 
(simple tensiometers would require many installation holes within a small 
area). The tensiometers are used to ensure that a steady state is reached 
during the infiltration measurement, to estimate the lateral flow 
components, to determine vertical infiltration rates and to estimate the 
saturated hydraulic conductivities of the different soil horizons. After the 
steady-state parameters are obtained, a plastic sheet is placed on the soil 
to prevent evaporation, and the soil water tensions are periodically 
recorded during the subsequent drainage period. From these, the 
unsaturated hydraulic conductivity as a function of soil water tension K(\y) 
is obtained. The unsaturated hydraulic conductivity can also be measured 
as a function of soil water content ^(9) with multiple TDR probes. This 

Soil Water 229 

can be advantageous because the hysteresis ofK(Q) is less than that of K(\y) 
(Mualem, 1986). The use of neutron probes or TDR in access tubes to 
measure the soil water content is also possible, although the installation of 
the wider access tubes in the small ring area can disturb the soil. 

Rainfall simulators or sprinkler infiltrometers 

These are arrangements of droppers or nozzles supplying the soil surface 
with a uniform inflow of water such that the flux density at the soil surface 
is kept at a constant level (Peterson and Bubenzer, 1986; Kutilek and 
Nielsen, 1994). The principal advantage in comparison with double-ring 
infiltrometers is abetter simulation of the infiltration process under natural 
rainfall conditions (Bouwer, 1986; Wallace, 1996). The relatively large 
sample area of these infiltrometers should increase the representativeness 
of the measurements. Small systems are available for small-scale 
applications in agroforestry plots (Asseline and Valentin, 1978; Mathieu 
and Pieltain, 1998). 

Borehole in filtro meter, well permeameter or Cuelph permeameter 

This is a subsurface technique consisting of a mariotte flask that is lowered 
into a well or borehole augered into unsaturated soil to the desired depth 
of measurement. The mariotte flask maintains a constant depth of water 
in the hole as a means of measuring the rate of water flow into the 
surrounding unsaturated soil (Elrick and Reynolds, 1992). 

The hole should be excavated using a screw-type or bucket auger. It 
should be cylindrical and have a flat bottom at least 20 cm above the water 
table (Reynolds, 1993). Specific factors that can affect the accuracy of the 
measurements with well permeameters include smearing, remoulding and 
compaction of the well surfaces during augering and gradual siltation of 
the well (Elrick and Reynolds, 1992; Reynolds, 1993). On clayey soils, 
scratching the bottom and side of the hole with a sharp-pointed instrument 
or metal brush normally alleviates the smearing and compaction caused 
during augering. On sandy soils, the collapse of the well during the 
measurement is the principal constraint to the use of this technique, 
whereas in silty soils the gradual siltation (clogging of the soil pores with 
silt particles) can be partially alleviated by protecting the bottom of the 
hole with coarse sand or fine gravel (Reynolds, 1993). Solar heating of the 
head space in the water reservoir can also affect the accuracy of the 
mariotte-based well permeameter, but this can be alleviated by performing 
the measurements under a tent or big umbrella. A further discussion of 
the method and procedures for calculating hydraulic parameters are given 
by Reynolds (1993), Elrick and Reynolds (1992) and Stephens (1996). 

230 W.C. Teixera ef al. 

Unsaturated hydraulic conductivity 

Most of the water movements above the water table in the field, including 
water and nutrient flux to plant roots, rainfall infiltration and leaching of 
nutrients through the soil profile, occur while the soil is unsaturated and 
are thus controlled by the unsaturated hydraulic conductivity. Unsaturated 
hydraulic conductivity can also be used to characterize changes in soil 
structure (Watson and Luxmoore, 1986; Wilson and Luxmoore, 1988; 
Ankeny et al., 1990; White et al, 1992), and this is an application of great 
relevance for agroforestry (see Fig. 11.4). Field measurements of 
unsaturated hydraulic conductivity are generally preferable to laboratory 
measurements if the site is sufficiently accessible, reasonably homogeneous, 
has a level topography, is not too stony and has predominantly vertical 
flow during drainage (Green et al., 1986). 

Tension in filtro meter or disc permeameter 

This infiltrometer consists of a porous baseplate, which establishes 
hydraulic continuity with the soil through a nylon membrane. It is 
connected to a mariotte reservoir that supplies the soil surface with water 
at a constant and regulated tension. Tension infiltrometers normally 
operate in the range of to -20 hPa. The data can be recorded manually 
or automatically (Ankeny, 1992; Elrick and Reynolds, 1992). 

Tension infiltrometers have been used to characterize near-saturated 
hydraulic properties, sorptivity, macroscopic capillary length, characteristic 
pore size (Ankeny, 1992; Reynolds, 1993; Stephens, 1996), soil structural 
conditions at the soil surface as a function of short-term variations in 
weather conditions (White and Perroux, 1989), effects of tillage practices 
(Ankeny et al., 1991; Messing and Jarvis, 1993) and root growth (White et 
al., 1992). The near-saturated hydraulic conductivity and sorptivity 
obtained using tension infiltrometry have been used to distinguish 
infiltration through macropores from that through the soil matrix (Watson 
and Luxmoore, 1986; Wilson and Luxmoore, 1988; Messing and Jarvis, 
1993). The method involves supplying water to the soil surface at different 
potentials to exclude a range of macropores from the flow, whose 
contribution to infiltration can thus be quantified. 

A good soil surface preparation and hence good contact between the 
infiltration disc and the soil is essential. The supply membrane must be 
visible during the infiltration to permit examination for air leaks (Perroux 
and White, 1988). In older models the water tower was mounted on the 
infiltrometer disc, but these are now constructed separately, which makes 
the measurement more stable and accurate under windy conditions and 
avoids collapse of soil structure near saturation. For measurements near 
saturation the infiltration disc must be level; otherwise the potential varies 

Soil Water 231 

across the supply surface. Close to saturation, small changes in soil tension 
lead to dramatic changes of the infiltration rate and unacceptable errors 
in structured soils. The contact material can have a large influence on the 
pore water pressure head and hydraulic head gradient at the soil surface, 
and this can have a substantial impact on the validity of tension 
infiltrometer results (Reynolds and Zebchuk, 1996). A shortcoming of the 
technique is the time necessary to reach steady flow at low tensions in 
clayey soils, which can make manual recording impracticable. In this 
situation an automated infiltrometer should be used. As for Guelph 
permeameters, solar heating of the head space in the mariotte reservoir 
should be avoided by shading. 

According to Reynolds (1993), the main factors affecting the accuracy 
of tension infiltrometer measurements are soil heterogeneity, soil collapse 
under the infiltrometer during the measurement, inadequate or changing 
hydraulic contact between infiltrometer and soil, and the hydraulic 
properties and thickness of the contact material. 

Several methods have been reported for determining soil hydraulic 
parameters with infiltrometer data (Ankeny et al., 1991; Elrick and 
Reynolds, 1992; White et al., 1992; Logsdon and Jaynes, 1993; Simunek 
and van Genuchten, 1997). These methods differ in mathematical theory, 
type and number of infiltrometers used, and procedures of data collection. 
It is necessary to decide about the mathematical procedure for 
determining soil hydraulic parameters before starting the evaluations 
programme, because requirements concerning soil boundary parameters 
differ between methods. A modification of the double-ring method that 
allows the determination of unsaturated hydraulic parameters is discussed 
in the section on saturated hydraulic conductivity. 

Indirect methods or pedotransfer functions 

Field measurements of hydraulic properties are time consuming and 
labour intensive. Therefore, several attempts have been made to estimate 
these properties from readily available soil data such as particle size 
distribution, organic carbon and bulk density. An estimation method that 
describes soil water relationships based on other soil characteristics is called 
a pedo-transfer function (Tietje and Tapkenhinrichs, 1993). 

Empirical approaches such as linear and non-linear optimization and 
neural network techniques have been used to solve specific problems in 
the determination of soil properties. However, this kind of approach 
requires some skills in mathematics and statistical analysis. Semi-empirical 
approaches have been used to estimate hydraulic properties from pore 
size (Mualem, 1976) and particle size distribution (Arya et al., 1999). 

The most common indirect approaches are the so-called parametric 
models, including those by Campbell (1974) and van Genuchten (1980). 

232 W.C. Teixera ef al. 

Results to date have indicated that predictive models are relatively efficient 
for many coarse and medium-textured soils, but are unreliable for fine- 
textured and structured soils like structured clayey Oxisols (Tomasella and 
Hodnett, 1996). Eastham et al. (1988, 1994) and Chirwa etal. (1994a) have 
used these approaches in agroforestry research. 

11.5 Estimating Topsoil and Subsoil Water Use with Stable 
Isotopes (C. Schroth) 

The soil depth from which plants take up water is often of considerable 
interest, particularly where different species are growing together so that 
it determines the extent to which they compete with each other for the 
same water resources or complement each other by using water from 
different depths (see Section 11.1). Both hydrogen and oxygen possess 
stable, heavy isotopes (deuterium and ls O, respectively), and in some 
situations the isotopic composition of the xylem water of a plant can be 
used to determine the depth of water uptake in the soil. The technique 
has recently been reviewed by Pate and Dawson (1999), and further 
methodological details are provided by Jackson et al. (1995). 

A precondition for the application of this technique is that the isotopic 
composition of the soil water itself shows a change with depth that is 
sufficiently great to be detected reliably. Jackson et al. (1999) stressed the 
dynamic nature of the isotopic composition of the soil water and its depth 
distribution, which requires that site-specific, concurrent isotope profiles 
are established whenever xylem water values are analysed. This is partly 
due to the intra- and interannual variation of the isotopic composition of 
precipitation water, which showed a significant negative correlation 
between deuterium concentration and monthly rainfall at a Brazilian 
savanna site (Jackson et al., 1999). In seasonally dry climates, the 
evaporation of water from the soil surface during the dry season causes 
the enrichment of water in the upper soil layers with the heavy isotopes, 
deuterium and ls O, relative to water in deeper soil layers (Jackson et al., 
1995; D.M. Smith etal., 1997b). 

With the exception of mangroves, no isotopic fractionation occurs 
during water uptake and transport by plants. Consequently, the xylem 
water reflects the isotopic composition of the soil water at the depth of 
uptake, or the relative contribution of different water sources if water is 
taken up from more than one depth (e.g. groundwater and rain-derived 
water in the topsoil). However, isotopic fractionation does occur during 
transpiration, with preferential phase change from liquid to vapour for 
the lighter isotopes. As a consequence, the isotopic composition of the 
water in leaves and other transpiring tissues differs from that of the xylem 
and soil water. For woody plants, xylem tissue can be sampled from the 

Soil Water 233 

stem with an increment borer, or by mild vacuum extraction of twigs, 
branches and roots. Immature, unsuberized parts that are transpiring 
must be avoided (Jackson et al, 1995; D.M. Smith et al., 1997b; Pate and 
Dawson, 1999). For conifers, a method of extracting xylem water from logs 
with a displacement method has been described (Glavac et ai, 1990). For 
herbaceous species, non-chlorophyllous tissue at the interface between 
shoot and root has been used for water extraction (Dawson, 1993). 
D.M. Smith et al. (1997b) wrapped segments of the stem of millet 
(Pennisetum glaucum) plants in plastic film for at least 2 h before sampling 
them. In pot experiments, this method had been shown to avoid isotopic 
fractionation of xylem water by transpiration through the stem walls. For 
certain questions, the extraction and analysis of phloem sap can be of 
interest (Pate and Dawson, 1999). The isotopic composition of water 
samples can be determined with mass spectrometry, following either 
vacuum distillation (Jackson et al., 1995) or direct equilibration 
(Scrimgeour, 1995). The ratio is expressed as 8D or 8 ls O (see Box 4.2 on 
p. 90 for delta notation), relative to standard mean ocean water (SMOW) 
or Vienna-SMOW. 

Jackson et al. (1995) measured a gradient in the hydrogen isotope ratio 
of soil water with depth at the end of the dry season in the tropical forest 
of Barro Colorado Island, Panama. The gradient was most pronounced 
within the upper 30 cm of soil. Measuring the hydrogen ratio of the xylem 
water of different tree species, the authors concluded that evergreen 
species had access to water deeper in the soil than deciduous species, in 
agreement with the relatively high levels of physiological activity of the 
evergreen species during the dry season. Roupsard et al. (1999) studied 
the water relations of Faidherbia albida in a parkland agroforestry system 
in Burkina Faso. Comparing the oxygen isotope ratio of the xylem sap of 
the trees with that of the soil water at three times during the year, they 
concluded that at the beginning of the dry season, the trees took up water 
principally from the groundwater table at 7 m depth, whereas at the 
beginning of the rainy season, they probably took up water near the soil 
surface and may have competed here with the crops, although at a time 
when their transpiration was already strongly reduced by defoliation. At 
a third sampling occasion, at the end of the dry season, the depth of water 
uptake could not be clearly determined, as the isotope ratio in the soil 
water changed irregularly with depth and the ratio in the xylem sap could 
be explained with water uptake either from the subsoil (2-4 m) or from 
the groundwater table, or both. 

Similar problems were encountered by Le Roux et al. (1995) in a study 
on the spatial partitioning of soil water between shrubs and grasses in a 
humid savanna in Cote d' I voire. They could explain observed oxygen 
isotope ratios in the xylem water of the shrub Cussonia barteri during the 
dry season with water uptake either from the upper 30 cm of soil or from 

234 W.C. Teixera ef al. 

soil layers below 150 cm, which had a similar isotope ratio to the top soil. 
A later study of plant water relations showed that much of the water uptake 
by the shrub during the dry season probably occurred from the subsoil, 
to which the grasses had no access (Le Roux and Bariac, 1998). This 
indicates that, in studies of soil water use, isotopic techniques should be 
complemented by other methods, especially the monitoring of soil and 
plant water status. 

In a study using oxygen isotopes of water use by windbreaks in the 
Sahel, D.M. Smith et al. (1997b) demonstrated that the tree Azadirachta 
indica switched opportunistically during the year between the use of 
surface water and groundwater where the water table was not too deep, 
but took up water continuously from near the soil surface where the 
groundwater was not accessible, thereby increasing competition for water 
with crops. As an extension of the described methodology, Bishop and 
Dambrine (1995) used both the natural 8 ls O gradient in the soil and a 
tritium ( 3 H) tracer, which was sprayed on the soil surface to localize soil 
water uptake by conifers with improved precision. 

Hydraulic lift phenomena have also been studied with stable isotope 
techniques. Using the hydrogen isotope composition of soil and xylem 
water, Dawson (1993) showed that sugar maple trees (Acer saccharum) 
rooted through a hardpan in the subsoil and took up water from the 
groundwater layer, which was then released into the soil above the 
hardpan. As a consequence, the 8D values of the water in the topsoil 
changed in a systematic way with increasing distance from the trees, with 
ratios close to those of the groundwater in the proximity of the trees and 
ratios close to those of rainwater at 5 m distance from the trees. Analysis 
of the xylem water of understorey plants, which had no access to the 
groundwater, revealed that these took up substantial quantities of the 
hydraulically lifted water from the soil when growing close to the maple 
trees, and this apparently improved their water relations and growth. This 
study and the employed methodology are of obvious relevance for 
agroforestry situations with deep-rooted trees and associated, shallower- 
rooted crops. For further applications of the methodology see Pate and 
Dawson (1999). 

Chapter 1 2 
Root Systems 


Biological Dynamics of Forest Fragments Project, National Institute 
for Research in the Amazon (IN PA), CP 478, 6901 1-970 Manaus, 
AM, Brazil 

12.1 Synopsis 

Root systems generally receive less research attention in agroforestry than 
they deserve, largely because of methodological difficulties associated with 
their study, especially the tedious and time-consuming nature of data 
collection. Despite these difficulties, a number of studies on tree root 
distribution in agroforestry systems are now available and have led to a 
reappraisal of the popularly held belief that trees are generally deep- 
rooted and so do not compete severely with crops for water and nutrients 
in the topsoil (Jonsson et al., 1988). In fact, many trees do have deep roots, 
but they also have roots in the topsoil and acquire their nutrients and water 
where these are most available. In humid climates and during the wet 
months in seasonally dry climates, nutrients and water are usually most 
abundant and available in the topsoil, where tree roots often compete with 
crop roots for them (see Box 5.2 on p. 100). Such insights emphasize the 
need for a better understanding of the root ecology of tree-crop 
associations as a precondition for developing agroforestry practices that 
maximize beneficial interactions between trees, crops and soil, and 
minimize tree-crop competition. 

The tension between expected soil-improving effects of tree roots on 
the one hand and competition with crop roots on the other is a specific 
feature of agroforestry associations (Schroth, 1995). Some competition is 
probably inevitable for efficient use of the available soil resources by the 
tree-crop association as a whole, but too much competition can easily cause 
economic failure of a system (see Section 5.1). The balance between 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 235 

F.L. Sinclair) 

236 C.Schroth 

beneficial and competitive effects of trees is especially critical on shallow, 
nutrient-poor and periodically dry soils, where soil resources are most likely 
to limit growth. On such sites, root processes in agroforestry associations 
require careful optimization, using the range of techniques that are briefly 
outlined below. In contrast, on sites without severe water and nutrient 
constraints, farmers may show little concern for root interactions between 
associated species. For example, on sites with sufficient rainfall and 
fertilization in Costa Rica, farmers use fast-growing and competitive tree 
species such as Eucalyptus deglupta as shade trees in coffee because of their 
favourable crown form, rapid shade establishment and low pruning 
requirements (Tavares et at., 1999). On volcanic soils in East Java, farmers 
associate coffee with very fast-growing Paraserianthes falcataria trees (G. 
Schroth, personal observation). Root interactions between plants are also 
of less concern in fallow systems, where trees and crops are grown in 
sequence rather than simultaneously. Here, trees with large and potentially 
competitive root systems can be chosen for maximum soil improvement 
through soil organic matter enrichment, increase in soil porosity and deep 
nutrient capture without the danger of competition with crops growing in 
the same plot. However, in the small-scale patchwork of fallows and fields 
that is typical for some densely populated regions, root interactions may 
occur at the boundaries between adjacent crop and fallow plots and may 
affect a significant proportion of the cropped area (van Noordwijk, 1999). 
On most sites, either water or nutrients limit growth for at least part 
of the year, and the possibility of competition between trees and associated 
crops for these resources requires consideration. Farmers can optimize 
root-related (and most other) processes in agroforestry associations 
through a combination of species selection, system design and subsequent 
management of the components (see also Section 5.1). Species selection in 
the present context refers to the genetic determination of the root 
characteristics of a species, whereas system design and management 
address in part the environmental factors that intervene in the expression 
of these characteristics. These are crucial, because the tree and other 
species in agroforestry systems will rarely be selected only because of their 
root characteristics (although these may influence the decision), but rather 
because of more immediate economic criteria such as fruit or wood 
production, ability to substitute mineral fertilizer through nitrogen 
fixation, or suitable crown architecture for the shading of light-sensitive 
crops. System design and management with respect to root processes can 
provide ways to increase the flexibility that a land user has to choose plant 
species, even if these do not have ideal root characteristics for a certain 
type of agroforestry practice (Schroth, 1996). Extending an earlier concept 
by Bowen (1984), the systematic use of species selection, system design and 
management for optimizing root functions has been termed root 
management (Schroth, 1995). 

Root Systems 23 7 

Tree species selection according to root characteristics may consider 
criteria such as the ability to rapidly develop a deep root system for 
nutrient recycling from the subsoil in fallows (Jama et al., 1998), restricted 
lateral root development to reduce competition with crops (Laycock and 
Wood, 1963), a large root mass for physical improvement of compact 
subsoils (Schroth et al, 1996), or relationships between root characteristics 
and nutrient availability in the soil after fallow (Schroth et al, 1995b) (see 
also Fig. 5.2). Of particular interest for agroforestry are tree species that 
are not very competitive with crops but at the same time are soil- 
improving, a combination found, for example, with Gliricidia sepium 
(Schroth and Lehmann, 1995). 

System design should be based on what is expected from the roots of 
a tree in a specific system. Fast-growing trees planted at narrow spacing 
in fallows can rapidly explore subsoil horizons (Jama et al., 1998), but trees 
planted at wider spacing in tree-crop associations may be less efficient in 
this respect (see Section 8.1). If trees are needed for productive functions 
only, they may be planted in a different plot from the crops, but, if they 
are expected to improve the soil with their roots, these must have sufficient 
access to this soil. Planting trees on the plot boundary is a common strategy 
for reducing above- and below-ground interactions between trees and 
crops, although it may increase interactions with the neighbour's crops 
and create conflicts. An alternative method is to grow trees and crops in 
rotation as in improved fallows, but this limits the age that the trees may 
attain and, therefore, the range of products that they can produce. Fodder 
and firewood can often be produced in short fallows, but not timber, fruits 
and resins. 

There is a wide range of management measures that influence root 
distribution and root functions in agroforestry systems. Soil tillage destroys 
the tree roots in the topsoil shortly before sowing the crops and gives these 
a temporary advantage, although tree roots grow rapidly back into the 
ploughed soil (Schroth et al., 1995a). Tillage is a routine measure in systems 
with annual crops, with the exception of zero-tillage systems, so reducing 
root competition in this way does not cause extra costs. However, 
depending on the number and physical properties of tree roots in the 
topsoil, tillage can become more onerous and time-consuming in the 
presence of trees, especially when it is carried out with hand tools. Root 
pruning is a measure specifically designed to reduce root competition and 
may be carried out by deep ploughing along tree rows, possibly repeatedly 
during a cropping season, or by cutting off superficial tree roots with hand 
tools. The first of these options is restricted to mechanized agriculture and 
requires there to be sufficient space to pass between trees and crops with 
a plough, and the latter option is probably limited to a small number of 
trees. Competition for fertilizer may be reduced by restricting application 
to the zone of maximum root activity of the target crop or tree instead of 

238 C.Schroth 

broadcasting it over the whole plot area. This may be combined with 
exclusion of competing plants such as weeds and cover crops by clean- 
weeding in the respective zone, as is often done with tree crops. Clean- 
weeding can, however, have non-target effects such as reduced 
mineralization of soil nitrogen in the weeded area close to the trees 
(Schroth et al., 2000a, 2001c), so increasing the access of the tree to 
nutrients from fertilizer may lead to reduced access to soil-derived 
nitrogen. Additional fertilizer may reduce competition for the applied 
nutrients, but may also stimulate root growth, thereby increasing 
competition for other nutrients and water (Schroth, 1999). Shoot pruning 
of trees reduces their requirements for water and can be very effective 
in avoiding competition during drought periods (Smith et al., 1998). 
However, impacts are species specific (Jones et al, 1998) and pruning when 
trees are too young, or too severe and frequent pruning can induce shallow 
tree root systems (van Noordwijk et al., 1991a) and may negatively affect 
tree survival and, in legume trees, nitrogen fixation (see Chapter 13). 
Herbaceous plant species with competitive root systems tend to reduce the 
lateral spread and increase the depth of the root systems of trees with 
which they are associated (Yocum, 1937; Neves et al., 1998). It has recently 
been shown that simple grass strips can alter the root architecture of tree 
saplings by restricting their root growth in the direction of the grasses, a 
form of biological root pruning (Schaller et al., 1999). Whether this 
translates into a more favourable root architecture of older trees requires 
further study. More detailed accounts of the effects of different 
management practices on tree root systems have been given in recent 
reviews (Schroth, 1995, 1999; van Noordwijk et al., 1996). 

Obviously, most root management measures have certain costs in 
terms of additional fertilizer input, labour or reduced tree growth, and 
these have to be outweighed by the benefits, such as improved crop growth 
and higher soil fertility, for them to be sensible options for farmers. The 
manipulation of root processes is not usually the only objective of a design 
or management decision, so roots are often 'co-managed' with other system 
properties. For example, the principal objective of tillage may be to 
suppress weeds but it may also reduce tree root competition; tree spacing 
and shoot pruning affect shade and other microclimatic factors as well as 
root distribution and density; and grass strips may control soil erosion and 
runoff and may produce fodder in addition to their potential effect on tree 
root distribution. In northern Nigeria, branch wood from trees was 
valuable because the demand for firewood was high, so tree pruning before 
crop planting controlled root competition while providing an economic 
return to the labour involved in pruning (Jones et al, 1998). 

Root Systems 23 9 

Research needs 

Agroforestry root studies have so far mainly concentrated on static analyses 
of tree root distribution, whereas dynamic root studies and process- 
oriented root research in agroforestry are still in their infancy. In the 
future, more efforts should go into the investigation of spatiotemporal 
patterns of root distribution and function, including the responses of root 
systems to management actions and the interactions between tree root 
systems and other soil biota such as fauna, microbes and other root 
systems. The following areas of root research are of particular interest. 

• Spatial and temporal patterns of water and nutrient uptake from the 
soil as influenced by lateral and vertical root extension, root intensity 
including root length density, number of root tips and mycorrhization, 
and soil characteristics including soil physical properties, acidity, and 
water and nutrient content. Studies in this field may help in making 
practical management decisions, for example, through identification 
of zones and periods of competition between associated species, 
leading to efficient fertilizer placement, or through the identification 
of the potential for nutrient recycling from the subsoil, as well as by 
providing parameters for models of water and nutrient cycling. 

• Effects of soil properties, either physical (compact horizons), chemical 
(acidity) or biological (pests, diseases, competing root systems), as well 
as management actions, such as tillage, fertilizer application and 
mulching, on root distribution. 

• Effects of roots on the soil, especially through carbon inputs from root 
exudation and turnover, rhizosphere processes, creation of 
macropores and interactions with soil fauna. 

• Methods of studying root distribution and functions in the field with 
little disturbance over extended time periods, including the in situ 
identification of the roots of associated plant species. 

Further discussions of root processes and research methodologies in 
agroforestry are provided in Chapter 7 (root effects on nutrient leaching), 
Chapter 8 (lateral and vertical redistribution of nutrients), Chapter 10 
(root effects on soil structure), Chapter 14 (mycorrhizas) and Chapter 15 
(rhizosphere processes). The most recent and comprehensive text on root 
methods is that by Smit et al. (2000), which supersedes Bohm's (1979) 
classic work on the subject. Other useful sources of information on 
methods for measuring root systems and processes include Mackie- 
Dawson and Atkinson (1991) and several chapters in Lassoie and Hinckley 

240 C. Schroth 

12.2 Methods for Studying Root Distribution 

The analysis of root distribution in the field provides a basis for the study 
of root interactions with the soil and between associated plant species. 
Specifically, the processes of nutrient pumping from the subsoil or lateral 
nutrient capture by trees, which have been discussed in Chapter 8, require 
that the trees have access via their root systems to either deep or laterally 
distant soil which contains relevant amounts of nutrients and where these 
nutrients are sufficiently mobile to be taken up (van Noordwijk et al. , 1996). 
An investigation into these processes should thus comprise a comparative 
assessment of root, nutrient and water distribution in the soil, either in the 
vertical or in the horizontal direction from trees. The results of such an 
assessment may also serve as a basis for more detailed studies, for example 
by applying tracers into nutrient-rich subsoil horizons to which tree roots 
have access and measuring their uptake into the above-ground biomass 
(see Section 8.2). Also, at least some qualitative observations on the root 
distribution of trees and crops should be made in most agroforestry 
experiments as a precaution against interferences between neighbouring 
plots through far-reaching, lateral roots, which may invalidate the 
experimental results (see Section 3.2). 

The principal methods for the analysis of root distribution in the field 
are the extraction of soil cores and the profile wall method. Excavation of 
representative sections of a plot can also be used, but this is much more 
labour intensive and is more appropriate when coarse roots need to be 
quantified, as in studies of total carbon accumulation in root systems 
(Akinnifesi et al., 1999). Excavations are usefully combined with allometric 
techniques, such as relating (coarse) root mass to proximal root diameter 
(the diameter of main roots close to the trunk) or stem diameter (van 
Noordwijk et al., 1996). There are also several indirect methods for 
studying the distribution of root activity, such as the analysis of water and 
nutrient distribution patterns in the soil (Box 8.1 and Section 11.2) or 
isotopic methods (see Sections 8.2 and 11.5). Estimating tree root 
distribution from architectural characteristics, such as by the application 
of fractal relationships or by estimating relative rooting depth of trees from 
the angles of main roots close to the stem, is still at an experimental stage 
(van Noordwijk et al, 1996; Ong et al, 1999). 

Soil coring 

Soil coring has the advantage of being less destructive than the preparation 
of profile walls, so that the same plot or tree can be used repeatedly for 
the same type of study, for example at the beginning and at the end of the 
growing season or during successive years. This allows the study of 

Root Systems 241 

dynamic root processes (see Section 12.4). After washing the roots from 
the soil cores, quantitative measurements of root mass or root length per 
unit soil volume can be made, and live and dead roots of different species 
can often be distinguished. Roots are usually washed from the soil over a 
sieve with 0.5 mm openings. Recent research with Grevillea robusta and 
maize roots indicated that this sieve size is adequate for studies of root 
mass, but that a 0.25 mm sieve may be required for accurate measurements 
of root length (Livesley et a!,., 1999). To avoid bias due to contamination 
of roots by mineral soil, root mass is expressed as ash-free dry weight 
(Anderson and Ingram, 1993) or after conversion into a common carbon 
content of 45% (Schroth and Zech, 1995b). 

Soil coring is most appropriate for root studies in superficial soil 
horizons. Coring equipment which allows the extraction of samples from 
depths greater than 1 m tends to be expensive and is not always available. 
In very stony or very hard soil, soil coring may be difficult or impossible, 
and the excavation of soil blocks of known volume can then be a suitable 
though more time-consuming substitute. For determining the volume of 
an irregularly shaped soil block, the hole can be filled with sand or lined 
with a plastic bag and filled with water. 

Soil coring devices are usually relatively small in diameter (2-10 cm), 
so that only a limited volume of soil is sampled with each core (Caldwell 
and Virginia, 1989). As a consequence, the number of samples required 
to obtain representative root data may be large, especially in deeper subsoil 
horizons, where tree and crop roots usually occur at low densities (Mackie- 
Dawson and Atkinson, 1991). Soil coring for the assessment of subsoil root 
distribution in agroforestry may be most useful in dense and relatively 
uniform plant stands such as planted fallows. In systems with individual 
trees and pronounced spatial patterns, such as parklands or shade tree 
systems, careful design of the sampling scheme is necessary to produce 
meaningful results. 

The identification of sampling points for soil coring should follow the 
same criteria as for soil fertility studies (see Section 3.4). As root distribution 
usually exhibits pronounced spatial patterns within agroforestry plots as a 
function of the position of the different plant species, random sampling 
from a whole plot is only appropriate in very dense and homogeneous 
stands. In more typical, heterogeneous agroforestry plots, either precise 
sampling positions are defined such as a number of fixed distances from a 
certain tree species, and replicate samples are collected from each position, 
or the plot is subdivided into more or less homogeneous strata from which 
random samples are taken (stratified random sampling, see Section 3.4). 
The first of these options gives more precise information about spatial 
patterns and can be used for geostatistical analysis or other interpolation 
methods, whereas the latter strategy gives more representative values for 
the plot as a whole without the need for spatial interpolation. 

242 C. Schroth 

If the variability of the root data from individual samples is known, 
the number of replicate samples can be determined as described in Section 
3.2. For objectives that require high precision in the measurement of root 
parameters, the necessary sample sizes will often be large, due to the large 
variability typically found in root studies. The effort for sample processing 
can be reduced drastically by bulking the replicate samples from the same 
position, homogenizing them (on a level surface, not in a bucket) and 
collecting a representative subsample. This can be done by dividing the 
sample into four quarters and retaining one quarter, which is again 
divided into four quarters and so on. The total sample and the subsample 
are weighed to determine the volume of the subsample from the known 
volume of the total sample. Tests with soil from a groundnut field showed 
that the root data from subsamples of 5-10% were sufficient to be 
representative of the total sample (Schroth and Kolbe, 1994). 

Profile wall method 

In comparison with root coring, the profile wall method has the advantage 
that no specialized equipment is required and that the roots are studied 
in their natural position. A trench is dug to a suitable depth, and the 
distribution of tree and other roots is studied on the trench wall, where 
roots intersecting the wall have been exposed. The roots can be counted 
manually, either in situ or in the laboratory after being traced on to 
transparent plastic film, or the profiles may be filmed with a video camera 
and the roots quantified by image analysis. This sometimes requires 
previous coloration of the roots to increase their contrast with the soil 
(Jorge, 1999). Profile walls give a good opportunity for studying the factors 
which influence root growth and distribution, such as soil cracks, 
earthworm channels or discontinuities in the sedimentation, or the effect 
of the roots on soil structure. 

A significant drawback, however, with profile walls is that, even when 
applying great care in the preparation of the wall, many fine roots are lost, 
especially in clayey soils. For this reason, counting the root ends 
intersecting with the wall should not be seen as a quantitative method for 
the evaluation of root distribution without careful calibration against soil 
cores collected from different depths. In agroforestry associations, it could 
even produce misleading results because the finer roots of herbaceous 
species are much more likely to be lost during the preparation of the 
profiles than the coarser tree roots. There are also pronounced differences 
between tree species concerning the mechanical resistance of their fine 
roots and thus the likelihood of root loss. Furthermore, live and dead roots 
are difficult to distinguish in the profile wall, especially when the soil is 
dry. To avoid these difficulties, it is often advisable to quantify only coarse 

Profile width (cm) 
20 40 60 80 100 20 40 60 80 100 20 40 60 80 100 


,,' , .,' .. ^TV 


i -''..'..iV ,•*•, • ' ,*, 

50 - 

• ., i < V *••• s. 






100 - 


i' i • i 


i i ' i • 


.' i • ■ ' •* , 


,' ,,»• 

.i . ' ■ i '■ ( r . 

150 - 

1 ' m, ' ' 


1 »i i ' ' ' 

• /•»• •• • • I* 

1 m tree distance 

2.5 m tree distance 

4 m tree distance 

Fig. 1 2.1 . Distribution of primary and secondary roots (dots and vertical lines, respectively) of oil palm (Elaeis guineensis) in a 
monoculture plantation in central Amazonia. The roots were marked on transparent sheets on profile walls at three tree distances. For 
roots < 2 mm see Table 12.1 (reproduced with permission from Schroth et ai, 2000a). 

244 C. Schroth 

Table 12.1. Fine root length density of oil palm (Elaeis guineensis) at three tree 
distances in central Amazonia as measured with soil coring (roots <2 mm, in cm 

cm -3 ). For coarse roots see Fig. 12.1 . a 

_ ., , , Distance from the tree 
Soil depth 

(cm) 1 m 2.5 m 

4 m 

















2.8 1.3 

0.8 1.7 

1.1 1.4 

1.0 0.6 

0.2 0.2 

0.2 0.1 

Reproduced with permission from Schroth et al. (2000a) 

roots in the profile and to collect volumetric samples from different depths 
in the wall with a core or monolith sampler for the quantitative extraction 
of fine roots (Fig. 12.1, Table 12.1). Other disadvantages of the profile wall 
method are its destructive nature and the fact that root data measured on 
soil samples taken from a single profile wall in a plot are likely to be less 
representative for the plot than samples collected from several spots by 
means of soil coring. 

Methods for horizontal root distribution 

In studies of the lateral redistribution of nutrients within a system or 
landscape through tree roots, the horizontal root distribution especially 
in superficial horizons is of interest. In the case of tree rows such as 
boundary plantings, shelterbelts and hedgerows with close within-row 
spacing, the systematic pattern of tree root distribution is effectively 
reduced from three to two dimensions. Soil coring can then be an efficient 
strategy for studying root distribution. Soil cores are collected in lines 
parallel to the tree rows, and the cores from the same distance are mixed 
and subsampled for root extraction (Schroth et al., 1995a). Before the 
sample collection, superficial trenches can be opened at different distances 
from the trees to obtain an idea of the tree root density to be expected and 
to assist in the definition of the exact sampling design (distances, depth 
increments, number of sampling points per distance and depth as 
influenced by root density and heterogeneity). 

For individual trees, as in parklands and savannas, the tree root 
density may rapidly become too low and too patchy with increasing 

Root Systems 245 

distance from the trees to make soil coring an efficient strategy. Trenches 
can be dug either at increasing distances from a tree or as spiral trenches 
originating from the tree (Huguet, 1973), and the tree roots counted on 
the profile wall. Alternatively, or in addition, superficial coarse roots can 
be followed from the tree by successive excavation. Special attention should 
be given to patches of increased nutrient and water availability, such as 
ant and termite mounds, depressions receiving run-on water and eroded 
soil, and decomposing biomass (Mordelet et al. , 1996). 

In studies of vertical and horizontal nutrient redistribution by tree 
roots, soil analyses for nutrients should accompany the root studies. The 
samples can either be collected separately or subsamples from the 
volumetric samples collected for the root studies can be used. The latter 
method is most appropriate if the roots are not extracted from the whole 
volumetric sample, but only from a subsample. Nutrient uptake from the 
subsoil is most likely when the top soil is dry but the subsoil still moist. For 
this reason, monitoring the soil water content at different soil depths 
during wet and dry seasons is useful in studies of deep nutrient capture 
in seasonally dry climates. 

12.3 Distinction of Roots of Different Species 

In associations of several plant species, it is often of interest to know to 
which species certain roots belong. In many cases, the roots of a plant 
species are sufficiently distinct from those of all other species present at a 
site to be identified either in the profile wall or later in a Petri-dish under 
magnification. Typical criteria used are colour, diameter of the finest roots, 
surface properties, branching patterns and nodulation. In quantitative 
studies, care must be taken that all age classes of roots can be assigned to 
a species, as certain characteristics may not be equally visible in old and 
new roots. Sometimes it is not possible to distinguish the roots of every 
species, but it may still be possible to identify those of species groups, for 
example, to distinguish palms from dicotyledonous trees or legumes from 
non-legumes. Such distinctions may still be very useful in the 
interpretation of the data. The identification of species in the dead root 
fraction is often impossible. In the profile wall, some characteristics become 
less visible when the soil and the roots dry out, which can be avoided by 
shading and wetting the profile and in seasonally dry climates by carrying 
out such studies in the wet season. 

In some cases, the biomass (including the roots) of associated species 
may differ in their contents of certain stable isotopes, so that the analysis 
of these isotopes in the root biomass of the individual species and in a 
mixed root sample can be used to determine the proportions of roots of 
different species in the sample. The proportions are obtained on a mass 

246 C. Schroth 

basis and not on a root length basis. This is important, as species can vary 
considerably in the mass/length ratio of their roots (Caldwell and Virginia, 

Plants following either the C 3 or the C 4 photosynthetic pathway differ 
in their 13 C/ 12 C ratio, which can therefore be used to determine the 
contribution of each group to a root sample collected from an association 
(Svejcar and Boutton, 1985) (see Box 4.2 on p. 90). Lehmann et al. (1998a) 
used this approach in an agroforestry system with Acacia saligna and 
sorghum (Sorghum bicolor). 

In certain cases this method can be applied even if all plants at a site 
use the same photosynthetic pathway. In forests, the 8 13 C values of the 
plant biomass gradually increase with plant size from understorey plants 
to canopy trees due to differences in stomatal opening and 13 C/ 12 C ratios 
of the carbon dioxide at different heights above the ground. Consequently, 
roots from tall trees can be expected to have higher 8 13 C values than roots 
from smaller plants. This approach has been used to analyse the relative 
contribution of trees and understorey vegetation to the root mass in the 
soil down to 4 m depth in an Amazonian forest (Sternberg et al., 1998), 
and it could probably be used in a similar way in agroforestry systems with 
a forest-like structure, such as forest gardens. Attention should be given 
to the observation of significant differences between the 8 13 C values of roots 
and corresponding stems (Medina et al., 1991). 

Nitrogen4ixing species often differ from non4ixing species in their 
15 N/ 14 N ratio (expressed as S 15 N, see Chapter 13), so that this ratio could 
also be used to determine the contribution to mixed root samples of species 
in an association (Caldwell and Virginia, 1989). However, the differences 
are smaller than for the 13 C/ 12 C ratio and the ratios may overlap between 
nodulated and non-nodulated species (Guehl et al, 1998). The 15 N contents 
of plants are influenced by factors such as mycorrhization, drought and 
plant nitrogen status (Handley et al., 1999). Furthermore, there is 
considerable variability in space and time, the reasons for which include 
change in the 15 N content of soil with depth and landscape position as well 
as seasonal variations of nitrogen fixation (Handley et al, 1994). 

12.4 Methods for Root Dynamics, Production and Turnover 

Root dynamics is a summary expression for temporal changes of root 
properties such as their mass, length and relative distribution in topsoil 
and subsoil. In comparison with static root data such as those obtained 
from a single measurement of root distribution, information on root 
dynamics is often of considerable additional value for the interpretation 
of agroforestry experiments. For example, in improved fallow systems, the 
ability of the trees to recycle leached nutrients from the subsoil is a function 

Root Systems 247 

of the progressive colonization of the subsoil by tree roots (Jama et al., 
1998), and the speed of tree root growth in the subsoil may be one of the 
factors that determine the minimum time under fallow needed for effective 
improvement of topsoil fertility. In simultaneous agroforestry systems, the 
root dynamics of associated species may influence their below-ground 
interactions: if phases of maximum root growth occur at the same time for 
different species, below-ground competition is likely to be high, whereas 
a certain lag in these phases for associated plants may indicate a degree of 
complementarity in the use of soil resources (Campbell et al., 1994; Munoz 
and Beer, 2001). The analysis of temporal patterns of plant root growth 
can also show whether phases of increased root growth coincide or are out 
of synchrony with phases of increased nutrient availability in the soil, such 
as from nitrogen mineralization or fertilizer application, and this may 
influence the efficiency with which nutrients are taken up (see Section 6. 1). 
Root dynamics may also reflect the response of plants to management 
actions, such as shoot pruning (Nygren and Ramirez, 1995; Schroth and 
Zech, 1995a) or fertilizer application (Campbell et al., 1994), which may 
influence the competitive balance among associated species. 

Root turnover is a specific aspect of root dynamics referring to the 
fraction of a root system that is renovated during a certain time period 
(commonly a year) through the death of some roots and their replacement 
by new root growth. There is no universally accepted definition of root 
turnover. It has been defined by different authors as either the cumulative 
root production or the cumulative root mortality during a study period 
(which is numerically the same as long as live root mass is similar at the 
beginning and the end of a study period), divided by either the average, 
maximum or minimum root mass during this period (Gill and Jackson, 
2000). Defining root turnover as the cumulative root production divided 
by the maximum root mass has the advantage that it yields a turnover of 
1 for an annual crop that maintains all its roots until the end of the growing 
season (Gill and Jackson, 2000). However, using maximum (or minimum) 
instead of average root mass as a basis for the calculation has two 
disadvantages: (i) the generally greater uncertainty in the measurement 
of extreme than of average values, and (ii) the risk of missing the true 
maximum when sampling the roots. Therefore, calculating root turnover 
as cumulative production divided by average live root mass is likely to yield 
results that are less prone to experimental error. Average root mass should 
be calculated by suitable weighting according to what sampling intervals 
have been used. In any case, root turnover values should always be 
reported together with the calculation method used and should be 
accompanied by data on root production and mortality as well as average, 
maximum and minimum root mass during the study period to facilitate 
comparisons between studies. 

Root turnover may occur naturally or may be a response to 

248 C. Schroth 

management interventions such as soil tillage or shoot pruning (see Vogt 
and Bloomfield, 1991, and Eissenstat et al., 2000, for discussions of the 
mechanisms involved in natural root senescence and turnover). Root 
production and turnover are of interest in agroforestry studies because 
the investments of carbon and nutrients into plant root systems affect the 
resources that are available for above-ground growth and production. 
Furthermore, root turnover is a mechanism through which plants 
replenish the soil with organic matter, supply carbon and energy to the 
soil biomass and release nutrients into the soil. 

Several methods are available for studying root dynamics in agroforestry 
systems. Some of these may be used to obtain quantitative or 
semiquantitative information on root production and turnover, whereas 
others are useful for qualitative purposes. Quantitative methods tend to be 
very demanding in time. Also, root studies are generally characterized by 
relatively large variability, which is partly caused by the spatiotemporal 
heterogeneity of the soil environment and its usually unknown effect on the 
root systems of different plants. Careful definition of research objectives and 
corresponding selection of methods in accord with the available resources, 
especially labour, are particularly important in research on root dynamics. 
A recent review of research methods can be found in Vogt et al. (1998). 

Sequential soil coring 

This is a frequently used technique in research on root dynamics. It 
consists of the repeated collection of volumetric soil samples in the field, 
usually with a cylindrical soil corer, from which the roots are extracted by 
washing over a sieve, then separated into live and dead roots, diameter 
classes and plant species as required, and quantified in terms of their ash- 
free dry weight or length (see Sections 12.2 and 12.3). As mentioned 
before, the method is only suitable for relatively fine roots (smaller than 
about 5-10 mm), but these are also the most dynamic root fraction. 
Sequential coring can provide a quantitative picture of fine root systems 
at different sampling times, which may describe seasonal or phenological 
patterns and be related to water and nutrient uptake (Kummerow et al, 
1982; Singh and Srivastava, 1985; Schroth and Zech, 1995a; Lehmann et 
al. , 1998a; Jose et al., 2000). Sequential soil coring data have been used for 
calculating root production and turnover in agroforestry by Cuenca et al. 
(1983), Schroth and Zech (1995b) and Lehmann and Zech (1998). There 
are also several root production studies from tropical forests using 
sequential soil coring (Kummerow et al., 1990) or sequential excavation of 
soil blocks (Singh and Singh, 1981; Srivastava et al., 1986; Khiewtam and 
Ramakrishnan, 1993). Dynamics of mycorrhizal fungi have also been 
studied with this method (Fogel and Hunt, 1983). 

Root Systems 249 

The collection of soil cores should be timed so as to include expected 
minima and maxima as well as critical situations in the root development 
of the system and its components. For example, in a system with trees and 
annual crops, samples may be collected before sowing the crop (to 
characterize the tree-only situation), during early crop development (when 
the soil may be dominated by tree roots), during maximum root 
development of the crop (when the crop roots may be more abundant than 
the tree roots), and after crop harvest when the crop roots have died and 
are decomposing in the soil. In a system where trees are pruned as part 
of their management, appropriate sampling times may be before the first 
pruning at the onset of the rainy season (presumably maximum tree root 
development) and then at monthly or bi-monthly intervals to check for 
root decline caused by the pruning treatment. In systems with perennial 
crops, such as coffee or cocoa with shade trees, the most pronounced 
contrasts would be expected between samples collected during the wet and 
the dry season (Cuenca et al., 1983), or between samples collected during 
vigorous vegetative growth and when the crop is weakened by 
reproduction at the end of the harvest season. 

To increase the probability that temporal changes of root properties 
can be seen in the data, the influence of spatial heterogeneity should be 
reduced by carefully standardizing the sampling protocol. For example, 
samples can always be taken from the same distances or distance ranges 
from a tree row or a number of shade trees, using the same trees for each 
sampling. The sampling depth is best defined on the basis of preliminary 
investigations of vertical root distribution in the field, taking into 
consideration that the inclusion of subsoil samples may greatly increase 
the effort required for sample collection in the field, although the time 
required for sample processing is usually much less for subsoil than for 
topsoil samples. Pooling several soil cores from similar sampling positions 
and depths and subsampling for root extraction is an efficient way for 
reducing the processing time (Schroth and Kolbe, 1994; Section 12.2) and 
may often be the only way to cope with the large number of samples 
collected during studies of root dynamics. 

Estimating fine root production and turnover 

Several methods have been developed for estimating fine root production 
and turnover from sequential soil coring data. These calculations are based 
on the following equations (Publicover and Vogt, 1993; Vogt et a!,., 1998): 

LFR (2 = LFR„ + P - M (12.1) 

DFR <2 = DFR„ + M - D (12.2) 


C. Schroth 

where LFR and DFR are the biomass of live and dead fine roots, 
respectively, t is time period, and P, M and D are production, mortality 
and disappearance, respectively, between times t t and t r These equations 
can be combined to yield the basic equation of net primary production of 
fine roots (P): 





To calculate total below-ground carbon allocation, terms for carbon losses 
such as exudation, respiration and allocation to mycorrhizal fungi would 
have to be added to root production (Vogt et al., 1998). 

Calculations of root production make use of either one, two or all three 
of the terms in Eq. 12.3. The simplest method (max-min method; 
Publicover and Vogt, 1993) consists of the summation of increases in LFR 
between sampling dates. If the data show a single maximum and 
minimum, as, for example, with an annual crop or perennial vegetation 
in a strongly seasonal climate, then production is calculated by subtracting 
the minimum from the maximum LFR. Implicitly, this method assumes 
that root growth and death are separated in time and, if both processes 
occur simultaneously, production tends to be underestimated (but see 
discussion of other sources of error below). 

A second calculation technique, called the balancing transfers method 
(Fairley and Alexander, 1985), considers changes in both LFR and DFR 
according to a decision matrix (Table 12.2). However, it does not consider 
the decomposition of dead roots during a sampling interval and may, 
therefore, still underestimate total root production. In fact, for a 
hypothetical data set where LFR and DFR are constant through time, both 

Table 12.2. Decision matrix for estimating root production (P), mortality (M) and 
disappearance (D) from soil coring data of live and dead fine roots (LFR and 
DFR, respectively) according to the balancing transfers method. Annual 
estimates are made by summing the estimates from all sampling intervals within 
a year. 3 

Live roots 

Live roots 




Dead roots 








D = 

D = 


Dead roots 





M = -ALFR 

D = -ADFR 

D = -ALFR • 


Reproduced with permission from Fairley and Alexander (1985). 

Root Systems 251 

the max-min and the balancing transfers method would indicate zero 
production (Kurz and Kimmins, 1987), although this situation could in 
theory be the result of a perfect equilibrium between root production, 
mortality and decomposition. Root production could also be 
underestimated if sampling does not coincide with seasonal peaks and 
troughs in root mass (Publicover and Vogt, 1993). In comparisons of 
agroforestry with annual cropping systems, this bias is particularly 
problematic because root growth and mortality are likely to be more 
separated in time for annual plants than for trees, and root turnover could 
thus be more strongly underestimated in the agroforestry plots than in the 
agricultural control plots (Schroth and Zech, 1995b). 

The third and, theoretically, the most satisfactory approach to estimate 
root production is the compartment flow method (Santantonio and Grace, 
1987; Publicover and Vogt, 1993), which considers all three terms of 
Eq. 12.3. With this method, mortality (M) is calculated from changes in 
DFR and a decomposition constant k, assuming an exponential 
decomposition of the dead root mass. Production and decomposition are 
then calculated by transforming Eqs 12.1 and 12.2 above (Publicover and 
Vogt, 1993): 

M = ^[DFR,,-DFR„e<-*'>] / [1 - e<-*'»] (12.4) 

P = LFR (2 - LFR (1 + M (12.5) 

D = DFR„ - DFR S + M (12.6) 

The decomposition constant k is related to the disappearance rate DR (i.e. 
the fraction of dead roots that disappears during a certain time interval 
such as a month) by: 

/s=-ln(l-DR) (12.7) 

Equation 12.4 can be modified if root decomposition is known to differ 
from the simple exponential model. In theory, the compartment flow 
method allows accurate estimation of root production even when root 
growth, mortality and decomposition occur simultaneously. The principal 
problem with this approach is the need for an accurate decomposition rate 
for fine roots, which is difficult to obtain (see Section 6.3). Under 
agroforestry conditions, the decomposition rate would depend on the 
relative contribution of tree and crop roots, including its seasonal 
fluctuations, to the dead root pool, as these roots are likely to differ in their 
decomposition dynamics. Unfortunately, the method is relatively sensitive 
to errors in the decomposition rate. It may, therefore, be useful to provide 
production estimates for a range of decomposition rates (Schroth and 
Zech, 1995b). The method is also sensitive to misidentification of live and 
dead roots, which are often not easy to distinguish, especially for tree 
species with dark roots, and to loss of dead roots when processing the 

252 C.Schroth 

samples. However, modelling results indicate that the production estimates 
remain more accurate than those of the other two calculation methods 
unless the processing error becomes quite large (Publicover and Vogt, 
1993). Whenever possible, the compartment flow method should be used 
for calculating root production from sequential soil core data, possibly in 
combination with the balancing transfers method for comparison. 

As indicated above, the max-min and the balancing transfers methods 
have an inherent tendency to underestimate root production because they 
ignore the simultaneous occurrence of root production, mortality and 
decomposition. However, there is another source of error with these 
methods which may lead to an overestimation of root production. This 
error arises from the summation of random increases in root mass between 
sampling periods. It increases with increasing sampling frequency and is 
especially severe for sites where there are no pronounced seasonal changes 
in live and dead root mass. To avoid this error, it is recommended to use 
only statistically significant changes in root mass for production 
calculations with these two methods (Vogt et at., 1998). This is a serious 
limitation because it requires the separate processing of many replicate 
root samples which could otherwise be pooled and subsampled to increase 
the precision of the measurement that can be achieved per unit processing 
time (Schroth and Kolbe, 1994). In some cases this problem can be 
circumvented because it is known, or at least it is very likely, that roots 
have turned over, even if differences between sampling dates are not 
significant at a given level of replication. This would be the case when tree 
roots show pronounced seasonal fluctuations in accord with weather or 
management actions such as pruning, in systems with annual crops whose 
roots necessarily die after the harvest (Schroth and Zech, 1995b), or when 
additional observations (e.g. with rhizotrons, see below) indicate root 
turnover between sampling dates (Vogt etal., 1998). However, where such 
evidence is not available and root data show no pronounced seasonal 
patterns, as for perennial crops or trees in climates without pronounced 
seasons, the max-min and the balancing transfers methods should not be 
used for production calculations. Modelling results indicate that the 
compartment flow method is not prone to this source of error (Publicover 
and Vogt, 1993), and it is therefore not necessary to restrict root 
production calculations to significant changes in live and dead root 
mass. The compartment flow method is ideally combined with the 
pooling-subsampling method to reduce processing time (Schroth and 
Kolbe, 1994). As indicated before, however, further research on root 
decomposition rates for different trees and crops as influenced by soil and 
climatic conditions and their mathematical description is required to make 
full use of the potential of this method. 

Root Systems 253 

Sequential excavation 

As soil coring is only suitable for sampling fine roots, other methods are 
required when coarse roots are also of interest, such as in studies of total 
carbon accumulation or progressive soil exploration by coarse lateral or 
vertical roots. Dynamics of coarse roots can be studied by sequential 
excavation of trees of different ages, which may be combined with soil 
coring for quantification of fine roots. Sequential excavation was used, for 
example, by Yocum (1937) for coarse and fine roots in his studies of apple 
tree root development as influenced by associated crops and soil 
management practices. After suitable calibration, allometric techniques or 
simply the root-shoot ratio can be used for estimating coarse root biomass 
of trees (van Noordwijk et al., 1996). These techniques have rarely been 
used in agroforestry because of the considerable work involved and the 
focus on fine roots in most studies. They could, however, become more 
relevant with increasing attention given to the potential of trees to increase 
carbon sequestration in agricultural systems. 

Root ingrowth method 

This method is best suited for comparative studies of root growth for 
different time periods, sites or experimental treatments, although the 
method has also been used for estimating root production in tropical 
forests (Jordan and Escalante, 1980; Kangas, 1992) and shaded tree crop 
plantations (Munoz and Beer, 200 1). A certain soil volume (usually a core, 
but sometimes a soil block) is removed and replaced by root-free soil in a 
mesh bag, so that root ingrowth during the incubation period into a 
defined soil volume can be determined. The mesh bag can be made of 
nylon or metal netting with an approximately 5 mm mesh (Steen, 1991; 
Munoz and Beer, 2001). The soil in the ingrowth core is either root-free 
soil from the same site, or a standard substrate such as sand or vermiculite. 
If soil from the site is used, live and dead roots are removed as far as 
possible by dry sieving. The use of a polyethylene sheet, charged with static 
electricity by rubbing, can help in the removal of fine root fragments 
(Munoz and Beer, 200 1). These procedures will not produce a completely 
root-free soil, so root residues should be extracted from a subsample for 
correction of the final data of root ingrowth (Persson, 1990). Incubation 
periods of 1-12 months have been used in humid tropical (Cuevas and 
Medina, 1988; Raich et al, 1994; Muhoz and Beer, 2001) and 
Mediterranean climates (Fabiao et al, 1985). Depending on the research 
objective, either all cores are collected after a fixed incubation period, or 
sets of cores are collected after increasingly long incubation periods (e.g. 
3, 6, 9 and 12 months) to study the progress of root ingrowth into the 

254 C.Schroth 

cores. In the absence of site- and species-specific experience with the 
method, the latter approach is probably preferable. With the ingrowth 
technique, Fabiao et al. (1985) studied the seasonal dynamics of root 
growth in Eucalyptus globulus plantations in Portugal, and Munoz and Beer 
(2001) studied root growth dynamics and turnover in shaded cocoa 
plantations in Costa Rica. 

The ingrowth method is well suited for studying the effect of 
experimental soil manipulations on root growth. For example, Cuevas and 
Medina (1988) and Raich et al. (1994) demonstrated limitation by certain 
nutrients for forests in Amazonia and Hawaii, respectively, through 
increased root ingrowth into cores with standard growth media to which 
these nutrients had been added. With a similar design, McGrath et al. 
(2001) could not clearly detect phosphorus deficiency in an Amazonian 
agroforestry system despite very low soil phosphorus availability. Steen 
and Hakansson (1987) demonstrated reduced root growth of annual crops 
into cores with increasing soil compaction. 

A disadvantage of the method is that, unlike sequential soil coring, it 
does not provide information on root distribution in the soil. If this 
information is required, the method has to be combined with another 
method, such as soil coring (Muhoz and Beer, 2001) or the profile wall 

If quantitative growth data are needed, a potential problem is the 
difficulty of creating conditions in the core which are similar to those in 
the surrounding soil with respect to factors such as soil structure (and thus 
water content), nutrient mineralization and the presence of competing 
roots. This is much easier to achieve in a recently tilled top soil of an 
agricultural field than in a plot under permanent vegetation. In 
agroforestry studies with several contrasting plant species, such as annual 
crops and trees, artificial growth conditions in the cores could affect plant 
species differently, thereby introducing bias. 

If ingrowth cores are used for measuring root production, information 
on live roots, dead roots and root decomposition should be taken into 
consideration, as outlined above for sequential soil coring (Steen, 1991). 
Several authors found reasonable agreement of production values 
obtained with the ingrowth and sequential soil coring methods (Persson, 
1983, for a boreal pine forest; Hansson and Andren, 1986, for a perennial 
grass ley; Symbula and Day, 1988, for a swamp forest), whereas others 
found substantial differences in fine root production (Neill, 1992, for a 
prairie marsh; Makkonen and Helmisaari, 1999, for a boreal pine forest). 
Until research in tropical agroforestry systems provides evidence that (or 
under what conditions) root production estimates from ingrowth cores are 
reliable, it is recommended that this technique is used for comparative 
purposes or in combination with other methods for measuring root 

Root Systems 255 


Minirhizotrons are a non-destructive method for the in situ observation 
and quantification of roots in the soil. They consist of a transparent access 
tube, such as a glass or acrylic tube, or a hole in the soil, through which 
images of roots growing at the interface with the soil can be obtained with 
a fibre optic probe, miniaturized video camera, or simply a mirror and a 
camera with a macro-lens (Mackie-Dawson and Atkinson, 1991; van 
Noordwijk et a!,., 1996). Johnson et al. (2001) provide a recent review of 
the technique. 

Minirhizotrons have been used for collecting data on root length 
distribution, root dynamics (growth, senescence and decomposition), root 
properties (branching patterns and suberization) and interactions of roots 
with other organisms (herbivory, parasitism and symbioses such as root 
nodules formed with rhizobia). Due to their non-destructive nature, they 
allow roots to be monitored through different seasons, wetting-drying 
cycles, phenological cycles of a plant, or in response to management 
interventions such as shoot pruning or fertilization (Ball-Coelho et al., 
1992; Katterer etal., 1995; van Noordwijk et al. , 1996; Price and Hendrick, 
1998; Vogt et al., 1998). If roots of different species can be distinguished, 
the technique allows comparative studies of root dynamics of associated 
trees and crops or pastures in agroforestry associations, which may give 
important information on below-ground interactions between species 
(Campbell et al., 1994). Often, however, it is not possible to distinguish 
roots of associated species (Hendrick and Pregitzer, 1996). Several software 
packages are available for analysing root data from minirhizotron 
observations (Vogt et al., 1998). 

When working with minirhizotrons, an important consideration is how 
to avoid artefacts due to altered growth conditions of the roots at the 
interface between minirhizotron and soil. In sandy, uniform, well-packed 
soil, the presence of rigid observation tubes does not seem to influence 
root growth (Vos and Groenwold, 1987; Ephrath et al., 1999), although 
preferential root growth at the soil-tube interface may occur in finer- 
textured soils. In compact soil, this may lead to deeper root growth along 
the tubes than in the bulk soil (Vos and Groenwold, 1987). To reduce the 
problem of gap formation at the interface between rigid observation tubes 
and soil, several types of inflatable minirhizotron tubes have been tested. 
Gijsman et al. (1991) described a system consisting of a rubber tube, actually 
a modified motorcycle inner tube, which exerts a small pressure on the 
surrounding soil when inflated, thereby avoiding gaps at the interface, and 
is removed when observations are made. Advantages include the better 
visibility of the roots, which are not obscured by an imperfectly transparent 
tube, and the elasticity of the tube, which is important in swelling and 
shrinking soils (Gijsman et al., 1991; van Noordwijk et al., 1996). 

256 C.Schroth 

Another factor to be considered is the installation angle of the tubes. 
Root growth along the side of minirhizotrons has been observed with 
vertically installed, rigid tubes. Sometimes, but not always, this can be 
avoided by installing the tubes at an angle of 45° to the soil surface (Mackie- 
Dawson and Atkinson, 1991). However, in studies of deep root systems, 
as with many trees, installation angles other than vertical are impractical 
(Ephrath et al., 1999). In a study with Acacia saligna and wheat in a sandy, 
disturbed soil, Ephrath et al. (1999) found no difference in root length 
density measured with tubes placed vertically or at an angle of 45° in the 
soil. As discussed above, larger effects of the installation angle would be 
expected in more clayey soils. In studies with shallow-rooted crops, 
horizontal installation of minirhizotrons may be advantageous because it 
leads to more images per soil depth, thereby reducing variability of mean 
values (Dubach and Russelle, 1995) and allowing the analysis of spatial 
patterns. Johnson et al. (2001) provide a detailed discussion of installation 
methods for rhizotron tubes and recommend a waiting period of 6-12 
months between installation and data collection in order to avoid artefacts 
caused by soil disturbance. 

Light can negatively affect root growth, even at low intensities, and so 
its entrance into the tubes needs to be avoided (Caldwell and Virginia, 
1989). Also, temperature effects should be avoided by insulating exposed 
ends of rhizotron tubes above the soil surface (Samson and Sinclair, 1994). 

The minirhizotron technique is well suited for comparative 
measurements of root dynamics in different seasons or experimental 
treatments. Repeated observations of the same root can provide 
information on root longevity and turnover. Johnson et al. (2001) 
recommended sampling intervals of 2 weeks or less in such studies in order 
to reduce underestimates of root turnover, which occur when roots appear 
and disappear between two sampling events. Difficulties have been 
encountered when attempting to transform minirhizotron readings into 
quantitative values of root length or root mass per unit bulk soil. 
Calibrations of minirhizotron observations against soil coring data 
produced relationships that varied with plant species and phase of crop 
development (Ball-Coelho et al., 1992; Merrill and Upchurch, 1994; 
Samson and Sinclair, 1994). Furthermore, for both herbaceous crops and 
trees, correlations between these methods tend to be poorest near the soil 
surface, where root length is often underestimated by the rhizotron 
technique (Ball-Coelho et al, 1992; Samson and Sinclair, 1994; Ephrath 
et al., 1999). This is problematic, because most roots are usually 
concentrated here. A theoretical calibration approach for estimating root 
length from minirhizotron data also produced contradictory results 
(Merrill and Upchurch, 1994). More encouraging results were recently 
obtained for estimating root mass of maize and two temperate tree species 
from minirhizotron observations of root surface area (Jose et al, 2001). 

Root Systems 257 

Despite these difficulties, the minirhizotron technique can be useful in the 
estimation of root production if the turnover values obtained from 
repeated observations of roots at different depths are combined with 
absolute root mass data as obtained, for example, with soil coring at one 
or several points in time. Studies that have used this approach have been 
reviewed by Hendrick and Pregitzer (1996). 

Nutrient fluxes associated with root turnover 

Nutrient fluxes through fine root systems of trees are often calculated by 
multiplying estimates of root production with nutrient concentrations in 
root samples, assuming that there is no retranslocation of nutrients from 
senescing roots (Burke and Raynal, 1994; Lehmann and Zech, 1998; 
Munoz and Beer, 2001). Available evidence on this topic is contradictory. 
A study by Nambiar (1987) suggested little translocation of nitrogen, 
phosphorus, potassium and magnesium from senescing fine roots ofPinus 
radiata, but other studies have indicated significant retranslocation of 
nitrogen and phosphorus from senescing roots of northern conifers (Meier 
et al., 1985; Ferrier and Alexander, 1991). A recent analysis of a large data 
set of nutrient concentrations in live and dead roots found no evidence 
for retranslocation of nitrogen, but indicated an average retranslocation 
of about 30% of phosphorus and potassium from senescing fine roots. For 
magnesium the average tissue concentration decreased by 25% between 
live and dead fine roots, but the difference was not significant (Gordon 
and Jackson, 2000). These results suggest that the turnover of phosphorus, 
potassium and perhaps magnesium in fine roots may be overestimated if 
calculated as indicated above, although estimates of nitrogen turnover in 
fine roots may in many cases be reliable. Further research is needed to 
confirm this. If roots of annual crops die after harvest or tree roots are 
sloughed off by tillage, retranslocation of nutrients obviously does not 

Chapter 13 

Biological Nitrogen Fixation 

K.E. Ciller 

Plant Production Systems, Department of Plant Sciences, 
Wageningen University, PO Box 430, 6700 AK Wageningen, The 

13.1 Synopsis 

Trees can increase nutrient inputs and reduce nutrient losses through a 
variety of processes. Biological nitrogen fixation is prominent among these 
as it is the only truly renewable source of nutrients in agroforestry or 
agriculture as a whole. Many of the trees used in agroforestry are legumes 
as the ability to fix nitrogen allows them to grow rapidly in nitrogen- 
depleted soils. Thus one of the principal roles of nitrogen-fixing trees in 
agroforestry is to improve soil fertility, athough many legume trees have 
multiple uses for fodder, fuelwood, fruits and timber. Legume trees are 
thus found in almost all types of agroforestry systems: as shade trees in 
perennial crops, in improved fallows, in hedgerow intercropping systems, 
as erosion barriers, live fences, isolated trees in parklands and fodder trees. 
Not all legumes can nodulate and fix nitrogen, and the taxonomy of 
the Leguminosae is a good guide to which legume trees are nitrogen fixers. 
Nitrogen-fixing symbioses are also formed between non-legume trees and 
actinomycetes (Frankia spp.), which are termed actinorhizal symbioses 
(Giller, 2001). The most important of these are the Casuarina spp. from 
Australasia, which are used for soil stabilization, as windbreaks and for 
poles and fuelwood throughout the tropics. The actinorhizal symbioses 
have been less studied in agroforestry than leguminous trees, but the 
methods outlined here can equally well be applied to the study of either 
type of symbiosis. A full list of nitrogen-fixing trees also includes larger 
species of cycads, which form symbioses in their leaf axils with blue-green 
algae (Cyanobacteria), but these are of no importance in agroforestry. 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 259 

F.L. Sinclair) 

260 K.E.Ciller 

Nodulation and nitrogen fixation in the Leguminosae 

The Leguminosae are a large and diverse family of flowering plants 
composed of some 18,000 species, ranging from huge, long-lived forest 
trees to tiny, annual, herbaceous forbs. All legumes have nitrogen-rich 
tissues, irrespective of whether they can fix nitrogen, and it is now thought 
that the nitrogen fixation ability evolved to satisfy the demand for nitrogen, 
rather than nitrogen fixation resulting in the high tissue concentrations 
of nitrogen, which had often been assumed earlier (see McKey, 1994; 
Giller, 1997). Those legumes, which are not able to fix nitrogen, also have 
a large tissue nitrogen demand, which they satisfy either by competing 
actively for soil nitrogen or by growing more slowly and conserving their 
nitrogen within the plant. A further major consequence of the large tissue 
concentrations of nitrogen in the Leguminosae is the wide variety of 
secondary metabolites which are produced to protect legume tissues from 
predation (Waterman, 1994). Among these secondary metabolites the 
polyphenols, and particularly the condensed tannins, play a major role in 
regulating nitrogen release from legume tissues in decomposition (see 
Sections 6.1 and 6.4) and during ruminant digestion (Giller, 1997). 

The Leguminosae are divided into three subfamilies, the Caesalpinioideae , 
the Mimosoideae and the Papilionoideae, and these subfamilies provide a 
useful guide as to whether the legumes are able to nodulate and fix 
nitrogen. The Caesalpinioideae , of which only 23% form nodules (de Faria 
et al., 1989; Sprent, 2001), are considered to be the most primitive group. 
Within the Caesalpinioideae, nodulation is restricted to a few tribes, the most 
notable of which is Cassieae, which appear to form a bridging group 
between the non-nodulating and nodulating legumes. Within this tribe all 
members of genera Cassia (sensu stricto) and Senna are unable to nodulate. 
The genus Senna includes several important agroforestry species, such as 
S. siamea and S. spectabilis, which have been widely tested in hedgerow 
intercropping and were earlier assigned to the genus Cassia (Irwin and 
Barneby, 1981). Closely related to the genera Cassia and Senna is the genus 
Chamaecrista, in which all the species examined nodulate and fix nitrogen. 
The range of nodule morphology within the genus Chamaecrista is striking, 
with some species having rather primitive nodule structures, in which the 
bacteria are maintained in persistent infection threads within the nodule 
cells, whereas in other species fully differentiated bacteroids are released 
within peribacteroid membranes in the host cell cytoplasm (Naisbitt et al. , 
1992). Persistent infection threads are also found in the only non-legume 
tree genus, Parasponia (Ulmaceae), which can form nodules and fix nitrogen 
with rhizobia (Trinick and Hadobas, 1988; Becking, 1992). 

Almost all legumes in the other two subfamilies, the Mimosoideae (90%) 
and the Papilionoideae (97%), can form nodules with rhizobia and fix 
nitrogen (de Faria et al., 1989). The exceptions in the Mimosoideae appear 

Biological Nitrogen Fixation 261 

to be cases where loss of nodulation and nitrogen-fixation ability has 
occurred in species which are found in very arid environments (e.g. some 
Acacia spp.), presumably due to the sensitivity of nitrogen fixation to 
drought (Odee and Sprent, 1992; Sprent, 1994). Members of the 
Papilionoideae which do not nodulate are largely those found in primitive 

Useful reference texts with information on nodulation and nitrogen 
fixation in legumes are Allen and Allen (1981) and Sprent (2001). 

Other sources of nitrogen fixation in agroforestry systems 

Apart from inputs from nitrogen-fixing trees, herbaceous cover crops/ 
green manure legumes and grain legumes are often important com- 
ponents of agroforestry systems and further information can be found in 
Giller (2001). The discovery of endophytic nitrogen-fixing bacteria in the 
tissues (particularly the vascular tissues) of a range of non-legume species, 
including oil palm (Elaeis guineensis), has led to claims that such bacteria 
may give substantial inputs of nitrogen. Direct evidence for substantial 
nitrogen inputs (—20 kg N ha _1 year 1 ) is strongest in sugar cane (Saccharum 
officinarum; e.g. Boddey, 1995), though the evidence is still controversial. 
Nitrogen inputs from heterotrophic, free-living nitrogen-fixing bacteria 
in soil are generally considered to be very limited and not of significance 
except over long time scales (Giller and Day, 1985; Vanderleyden, 1997; 
Giller, 2001). 

Environmental limitations to nitrogen fixation 

All environmental limitations that adversely affect plant growth and vigour 
also decrease amounts of nitrogen fixation in legumes, although the 
symbiosis is sometimes more sensitive to such constraints than other aspects 
of plant growth (Giller, 2001). Nitrogen fixation is sensitive to nutrient 
deficiencies, in particular phosphorus deficiency, which may restrict 
nodule formation if acute. Molybdenum deficiency influences nitrogen 
fixation directly as molybdenum is a component of the nitrogenase 
enzyme. Nitrogen fixation is thought to be more sensitive to drought stress 
than other processes, such as photosynthesis, although the evidence is 
somewhat equivocal (Sprent, 1984). A further interesting feature of the 
legume-rhizobium symbiosis is that the sensitivity to stress may be 
expressed through the bacteria, the legume host or the formation of the 
symbiosis itself. Legumes are invariably more sensitive to salinity than are 
rhizobia (Sprent, 1984), but rhizobia are more sensitive than their hosts to 
heavy metal pollution (Giller et al., 1998). The infection process itself 

262 K.E.Ciller 

appears to be particularly sensitive to calcium deficiency (Giller, 2001). 
Large differences in sensitivity to stresses such as aluminium toxicity in 
soil are found among rhizobial strains and among legume hosts. 

Effects of management on nodulation and nitrogen fixation 

Legume trees are often pruned severely in agroforestry systems to provide 
fodder or foliage for soil amendment. It is well established that defoliation 
causes nodule senescence (Witty and Minchin, 1988), and pruning or 
browsing of trees by animals causes temporary decreases in the rates of 
nitrogen fixation. Reestablishment of nitrogen fixation will depend on 
formation of new nodules, though this can sometimes be rapid as legumes 
often harbour latent infections in young roots, which can develop when 
nitrogen demand in the plant is large. However, in Erythrina poeppigiana 
pruning resulted in complete mortality of nodules and there was a lag of 
10 weeks before active nodules were re-formed (Nygren and Ramirez, 
1995). Similarly, pruning of Leucaena diversifolia resulted in drastic 
reduction of nodule activity (measured as acetylene reduction) for about 
3 months (Snoeck, 1995). Nygren (1995) highlighted the danger that 
nodulation might be completely suppressed if trees are pruned too 

Nodule senescence associated with defoliation of trees is a mechanism 
by which nitrogen is made available to plants growing in close proximity, 
though amounts of nitrogen made available in this way are likely to be 
relatively small (Rao and Giller, 1993; Nygren and Ramirez, 1995). Even 

Table 13.1. Mechanisms by which nitrogen from nitrogen-fixing trees can be 
made available to other plants. 3 

Rate of Likely importance 

Mechanism transfer as an N source 

Below ground 

Root and nodule senescence and mineralization Slow Major 

Rhizodeposition Rapid Minimal 

Transfer between roots by interconnected 
mycorrhizal hyphae Rapid Minimal 

Above ground 

Mineralization of severed or senesced plant material Slow Major 

Consumption by grazing animals or insects and 
return in excreta or as carcasses Slow/rapid Major 

Foliar leachates Rapid Moderate 

Transfer of ammonia to associated plants Rapid Minimal 

a Modified from Ledgard and Giller (1995). 

Biological Nitrogen Fixation 263 

Box 13.1. Rhizobia that nodulate legume trees. 

The rhizobia that nodulate legume trees are a sadly neglected area of study 
compared with the amount of research effort which has been directed to grain 
legumes. The taxonomy of rhizobia has developed rapidly since the advent of 
phylogenetic methods for bacterial classification based on sequence analysis of 
the 1 6S rRNA gene (Young and Haukka, 1 996). The older classification methods 
were largely based on the legume host range of the rhizobia. Although this 
approach was recognized to be severely flawed early on (Wilson, 1 944), the new 
classification reveals the true extent of such problems in that nodulation ability is 
closely related to the nodulation genes carried by the rhizobia, rather than their 
evolutionary similarity (Young and Haukka, 1996). The nodulation genes are 
generally carried on transmissible plasmids in fast-growing rhizobia, and in some 
cases on transmissible 'symbiotic islands' of chromosomal DNA (Sullivan and 
Ronson, 1 998). Rhizobia far apart on the phylogenetic tree may thus carry the 
same nodulation genes and have a very similar host range for nodulation and 
nitrogen fixation. 

Research in the last 1 years has revealed some surprising overlaps between 
rhizobia which can nodulate legume trees, herbaceous and grain legumes. The 
best documented case is that of Rhizobium sp. NCR234, which has been shown 
to effectively nodulate legumes from 1 1 2 genera, including members of the three 
different subfamilies of the Leguminosae (Pueppke and Broughton, 1 999). Other 
broad-host-range rhizobia that nodulate trees are Rhizobium tropici (Martfnez- 
Romero etal., 1 991 ), which nodulates Leucaena spp. and Phaseolus vulgaris, and 
indeed other rhizobia species such as R. etli, which were thought to be specific 
in their host range, have since been shown to nodulate a wider range of species 
including trees (Hernandez Lucas et al., 1 995). 

Slow-growing rhizobia are all members of a single genus, Brady rhizobium, 
whereas the fast-growing rhizobia have been split into several genera. Five genera 
of fast-growing rhizobia are currently recognized: Rhizobium, Azorhizobium, 
Sinorhizobium, Mesorhizobium and Allorhizobium, although this field is changing 
rapidly. Some trees are reported to nodulate only with slow-growing rhizobia 
(Dreyfus and Dommergues, 1 981 ) but the vast majority examined nodulate with 
fast-growing rhizobia. Of particular note are the rhizobia which form stem nodules; 
shrubby legumes of the genus Aeschynomene form stem nodules with 
photosynthetic slow-growing Bradyrhizobium strains, whereas Sesbania rostrata 
forms nodules with rhizobia classified into a separate genus, Azorhizobium 
caulinodans (Ciller, 2001 ). 

Recent studies of rhizobia that nodulate legume trees have resulted in the 
description of the new species Sinorhizobium terangae and Sinorhizobium saheli 
from nodules of Sesbania and Acacia spp. in Senegal (de Lajudie etal., 1 994) and 
Mesorhizobium plurifarium (de Lajudie ef al., 1 998), Sinorhizobium arboris and 
Sinorhizobium kostiense (Nick ef al., 1 999) from nodules of Prosopis and Acacia 
spp. in Sudan and Kenya. The classification has recently been substantially revised 
(see Young etal., 2001) 

264 K.E.Ciller 

with E. poeppigiana, which was profusely nodulated in the field, amounts 
of nitrogen released from complete senescence of the nodules were much 
less than 10 kg N ha _1 (Nygren and Ramirez, 1995). If the turnover of fine 
roots is included, the amounts of N recycled below ground are likely to be 
of considerable importance but accurate estimates are lacking. The major 
pathway by which nitrogen is made available for other plants is through 
recycling of above-ground litter either directly to the soil, or through 
manure of animals grazing on the trees (Table 13.1). 

Requirements for inoculation 

The success of many of the fast-growing legume trees in agroforestry has 
undoubtedly been due to their ability to nodulate and fix nitrogen in soils 
across the tropics. Unfortunately, it is almost impossible to generalize as 
to which species are likely to nodulate readily in soils where they have not 
been grown, although there are gradations of the degree of promiscuity 
or specificity among agroforestry trees (Bala and Giller, 2001). Bala (1999) 
hypothesized that the success of species such as Leucaena leucocephala and 
Gliricidia sepium could have been due to either: (i) their ability to nodulate 
promiscuously with a wide range of rhizobial types, or (ii) the fact that, 
although such trees may nodulate with a narrow range of rhizobial species, 
these species are ubiquitous throughout the tropics. To test these 
hypotheses he characterized rhizobial isolates from nodules of Leucaena 
leucocephala, Calliandra calothyrsus, Gliricidia sepium and Sesbania sesban 
grown in soils from Central and South America, West, East and southern 
Africa and South-east Asia. Rhizobial isolates were characterized on the 
basis of physiological tests, host range for nodulation and molecular typing 
methods (including 16S rRNA partial gene sequences). Bala (1999) found 
that rhizobial isolates nodulating Leucaena, Calliandra and Gliricidia were 
diverse, belonging to a wide range of strains of the genera Rhizobium, 
Mesorhizobium, Sinorhizobium and Allorhizobium/Agrobacterium. This supports 
the first hypothesis that the tree legumes nodulated promiscuously with a 
wide range of rhizobia. Further, these three tree legumes nodulated with 
many of the same isolates although there were isolates which nodulated 
with only one or two of the species. Sesbania sesban was an exception in that 
it nodulated with a largely separate range of isolates, which fell into the 
Rhizobium and Mesorhizobium genera. Even more surprising was the limited 
range of soils from southern Africa in which S. sesban nodulated, 
considering that it is naturally found in this region. This clearly indicated 
that S. sesban requires inoculation when sown for the first time into upland 
soils. Details of how rhizobial inoculants can be made and applied are given 
in the methods manuals cited below. 

Biological Nitrogen Fixation 265 

13.2 Microbiological Methods for Studying Rhizobia 

Methods for the isolation of rhizobia from nodules and authentication are 
simple and can be conducted in any laboratory which has the capacity for 
simple microbiology. There are a number of excellent methods manuals 
including Vincent (1970), Somesagaran and Hoben (1985) and Sylvester- 
Bradley and Kipe-Nolt (1988). Host-range studies provide information on 
the ability of individual isolates to nodulate and fix nitrogen with different 
legumes, and therefore can give some clues as to the phylogenetic 
relatedness of strains, but detailed strain characterization now relies on 
methods of molecular biology, which tends to restrict such activities. 

13.3 Simple Methods for Determining Whether a Legume is 
Fixing Nitrogen 

In the absence of access to isotope-based methods (see below) there are 
some simple approaches to assessing whether trees or crops are benefiting 
from nitrogen fixation. 

• The simplest and most obvious (though often overlooked) approach 
to determine whether a tree is fixing nitrogen is to excavate the root 
system and look for nodules. If nodules are found, then this is strong 
evidence that the trees are or have recently been fixing nitrogen. If 
no nodules are found this could be due to a number of reasons: that 
the root excavations are not complete and the nodules are only found 
in deeper horizons, or that the nodules readily become detached from 
the roots and lost during sampling, which is particularly problematic 
with some species (e.g. Cajanus cajan, Gliricidia sepium). 

• The ability of a tree to grow and accumulate nitrogen in infertile or 
skeletal soils may be evidence of nitrogen fixation. Sanginga et al. 
(1986) compared nitrogen accumulation in primings of inoculated and 
uninoculated Leucaena leucocephala and estimated that 224-274 kg N 
ha _1 came from nitrogen fixation. This approach to estimating 
nitrogen fixation, which is a simple and useful approach to estimating 
nitrogen benefits, is referred to as the nitrogen balance method. 
However, in many situations it is dangerous to rely on such evidence 
alone as tree roots might be accessing deep soil horizons richer in 
nitrogen (see Box 8.1 on p. 171), or there may be subsurface lateral 
flow of water from which the trees can scavenge nitrogen. A further 
development of this approach is to develop a whole nitrogen balance 
for the system, preferably comparing a system with nitrogen-fixing 
trees with a system with non-nitrogen-fixing trees. Such studies suffer 
from the large number of unknown processes such as gaseous losses, 

266 K.E.Ciller 

which are as difficult to estimate as nitrogen fixation. A primary 
advantage of agroforestry systems is that the trees are able to access 
nutrient sources which are not available to other plants, which in itself 
can cause problems in calculations of nitrogen balances if nitrogen is 
taken up from deep water tables. 

13.4 Isotope-based Methods for Measurement of Nitrogen 

15 N-isotope-based methods are perhaps the only means available for 
estimating amounts of nitrogen from nitrogen fixation in trees growing in 
the field, though such methods may have significant problems in their 
application. The two main variants of these methods are those involving 
addition of 15 N-labelled fertilizers (often called 15 N-isotope dilution 
methods) and those that rely on the natural 15 N -enrichments of soil (known 
as natural abundance methods). Most soils are naturally, slightly enriched 
with 15 N with 8 15 N values commonly of 2-16%c (the difference between the 
15 N content of the material and air expressed in parts per 1000): 

Sample — Reference 

8 15 N (in parts per 1000 or °/ 00 ) 



X1000 (13.1) 

where during chemical analysis R is the ratio of N 2 molecules derived from 
the plant material, which are composed of one 15 N and one 14 N atom to 
those composed of two 14 N atoms, i.e. 

15 N+ 14 N 
R= (13.2) 

14 N+ 14 N 

The basic principle of both methods is that nitrogen taken up from 
the soil is enriched with the heavy isotope 15 N compared with nitrogen 
absorbed from the atmosphere. A non-nitrogen-fixing reference plant is 
used to estimate the 15 N-enrichment of nitrogen available from the soil in 
both cases (Peoples et a!,., 1989). The major assumption (and problem) of 
these methods is that the 15 N -enrichment of the available soil nitrogen 
taken up by the reference plant is the same as that taken up by the 
nitrogen-fixing legume. This is in fact only true when there is a perfectly 
uniform 15 N-enrichment of plant available nitrogen in the soil with depth 
and time, or, if this is not the case, when the rooting depths and activity 
and the time course of nitrogen uptake of the reference and legume plants 
are identical (Witty, 1983). This assumption is rarely satisfied and therefore 
steps must be taken to limit the variability in 15 N -enrichment of the 
available soil nitrogen as far as possible (Witty and Giller, 1991). The use 
of direct measurements of the 15 N -enrichment of the available soil nitrogen 

Biological Nitrogen Fixation 267 

coupled with modelling have been recommended as an alternative to using 
reference plants, but such approaches require extensive sampling and 
analysis of extractable soil nitrogen (Chalk and Ladha, 1999) and are 
unlikely to be widely used. Much has been written on the problems of 
measuring nitrogen fixation using isotope approaches (e.g. Witty, 1983; 
Witty and Ciller, 1991; Chalk and Ladha, 1999). Problems encountered 
with herbaceous legumes when isotope methods are applied are magni- 
fied in studies with trees due to their much larger size and longevity 
(Boddey et al., 2000). The size of trees and lignification of plant parts 
presents additional problems of subsampling of plant tissues to give a 
representative sample for measurement of tissue nitrogen concentrations 
and 15 N-abundance. 

Box 13.2. Measuring nitrogen fixation under controlled 

There are remarkably few measurements of nitrogen fixation by nitrogen-fixing 
trees growing in the field. This is in no small part due to the technical problems 
that exist in the application of the available methods, and partly for this reason 
many measurements have been confined to studies under more controlled 
conditions with trees grown in pots. The major limitations of these pot studies is 
that they do not often provide the information we need to assess the role of nitrogen 
fixation in agroforestry systems. 

So what do we want to measure? Do we want to know which of the different 
provenances of Gliricidia sepium fix the most nitrogen after 6 weeks of growth in 
a glasshouse? Or do we wish to pose the question: are all rhizobial strains that 
belong to different species or genera equally efficient in nitrogen fixation with a 
particular tree species? If these are the types of questions raised, then experiments 
in pots are a valid approach, but we must further question what methods are 
required to address these questions. 

The principal reason for using the 15 N-isotope dilution method is to distinguish 
between nitrogen derived from the atmosphere and nitrogen derived from the soil. 
If experiments are conducted in pots, the supply of nitrogen from the soil can be 
carefully controlled, so there is little advantage in using the 15 N-isotope dilution 
method compared with simpler nitrogen balance methods. 

The major concern with such studies under controlled conditions in pots is 
that there is no guarantee that there is any correlation whatsoever between results 
found under controlled conditions in the glasshouse and the response of the same 
treatments in the field. This is largely due to the strong interactions between 
nodulation and nitrogen fixation and a wide range of environmental factors, which 
are described above. 

268 K.E.Ciller 

13.5 Estimating Nitrogen Fixation in Field Settings 

It can be argued that 15 N -isotope dilution using labelled fertilizers has no 
role to play in estimation of nitrogen fixation by trees, for the following 

• A wide variability in rooting patterns can be found even within a single 
provenance of a tree species, as soil chemical and physical properties 
exert a very strong influence on root distributions. As mentioned 
before it is difficult to ensure that 15 N is incorporated to depth and 
unlikely that non-fixing plants can be identified that have matching 
rooting patterns. 

• One of the major advantages of legume trees is that they can be deep- 
rooting, but this raises obvious problems in uniform 15 N-labelling of 
soil with depth. These are problems that also apply to measurement 
of nitrogen fixation in herbaceous legumes but are amplified in 
agroforestry systems because of the larger size and potentially deeper- 
rooting systems of the trees. 

• 15 N-isotopes are expensive. Although it can be argued that the cost of 
the isotopes is relatively small compared with other costs of field 
experimentation and mass spectrometric analysis of samples, the wide 
planting arrangements of trees in agroforestry systems demand the 
use of large 15 N-labelled microplots. There is, therefore, a strong 
temptation to try to cut costs by not including border rows in the 15 N- 
labelled microplots, but sampling the whole area to which 15 N -labelled 
fertilizers have been applied. Obviously, this breaks the basic rules of 
experimentation with 15 N -labelling (see, for example, Stumpe et al., 
1989). Given the extensive rooting systems which have often been 
observed in agroforestry trees, border rows are particularly important 
(see Section 3.2). 

This means that the 15 N natural abundance method is probably the 
only sensible choice for measuring nitrogen fixation by trees in the field, 
but again problems should not be underestimated. The 8 15 N of soil 
nitrogen may vary markedly with depth, although the extent to which this 
is a problem is debatable (Boddey et al, 2000). Some studies have found 
little change of 8 15 N in the plant-available nitrogen with depth, but there 
is considerable variation between sites so this should be investigated if 
possible. It is unlikely that a reference tree can be identified which will 
have an identical pattern of root activity with depth, or uptake with time, 
as that of the nitrogen-fixing tree under investigation. One approach to 
this problem is to use a range of potential reference plants. This was 
recommended for measurements of nitrogen fixation in grain legumes 
(Boddey et al., 1990) but is especially applicable for nitrogen-fixing trees. 
If most of the reference plants give S 15 N values close to the 8 15 N of the total 

Biological Nitrogen Fixation 269 

Kjeldahl nitrogen of the soil on which they are growing, but one reference 
plant consistently gives 8 15 N values significantly different from the rest, 
then that is a reasonable indication that it is an aberrant reference value 
and may be discarded. Such a situation was found when measuring 
nitrogen fixation in Faidherbia albida in Malawi where mango (Mangifera 
indica) consistently gave small 8 15 N values (2.7-4.8%c) compared with the 
soils (6.4-1 1.5%c) and other non-fixing trees on the same sites (6.0-10.0%o) 
(Phombeya, 1999). 

The reason(s) why certain plants tend to become depleted or enriched 
in 15 N relative to their sources of nitrogen are not fully understood but 
appear to be related to mycorrhizal colonization, drought and nitrogen 
deficiency (Hogberg and Alexander, 1995; Handley et al, 1996, 1999). 
Hogberg (1990) showed that ectomycorrhizal non-nodulating legume 
trees (Brachystegia and Julbernardia spp.) in miombo woodland of Tanzania 
were more depleted in 15 N than arbuscular mycorrhizal-nodulating or 
non-nodulating legume trees. Subsequent studies in Cameroon did not 
confirm that these differences were due to the type of mycorrhizal infection 
(Hogberg and Alexander, 1995). Other research suggests that 8 15 N 
depletion in trees may be related to fractionation during transfer from 
mycorrhizas and depletion of S 15 N in tree shoots may indeed be indicative 
of mycorrhizal status (Hobbie et al., 1999, 2000). Problems of this kind 
must be borne in mind if the 8 15 N method is to be used. As a general rule, 
other indicators of nitrogen fixation such as nitrogen accumulation and 
the ability of the tree to nodulate should be ascertained before this method 
is used. Application of the 15 N natural abundance method for measuring 
nitrogen fixation by trees is reviewed in detail by Boddey et al. (2000). 

13.6 Methods Based on Nitrogen Fixation Transport Products 

Nitrogen is taken up from soil principally in the form of nitrate, whereas 
fixed nitrogen is assimilated into amides or ureides depending on the host 
legume before transport to the shoots. The proportion of nitrogen derived 
from nitrogen fixation at any given time can be estimated based on the 
relative proportion of the nitrogen in the xylem sap present as ureides 
(Peoples et al., 1989). Samples of bleeding sap from cut shoots are taken 
for analysis of total nitrogen and ureide nitrogen. This method can also 
be applied by extraction of dried petioles and determination of the ureide 
content, which makes sampling easier (Peoples et al., 1989). 

Ureide production appears to be limited to legumes of the tribes 
Phaseoleae, Indigoferae and Desmodieae (Giller, 2001), so that the only 
agroforestry species confirmed to transport ureides are Cordariocalyx 
gyroides and Desmodium rensonii (Herridge et al., 1996). Flemingia macrophylla 
belongs to the tribe Phaseoleae and may therefore transport ureides (Giller, 

270 K.E.Ciller 

2001), but this has yet to be confirmed. Comparisons of field estimates of 
nitrogen fixation in Cordariocalyx with the 15 N natural abundance method 
showed close correspondence with estimates made using the ureide 
method (Peoples et al., 1996). Reports of ureide production in species 
within the tribe Robineae (such as Sesbania and Gliricidia) and the Mimoseae 
are almost certainly due to artefacts produced by coloured compounds 
during the analysis procedure. Attempts have been made to estimate 
nitrogen fixation in amide-exporting legumes using xylem sap analysis, 
but with less success (Peoples et al, 1989). 

13.7 Estimating Total Amounts of Nitrogen Fixation 

With all of the above methods, biomass estimates and tissue nitrogen 
analyses are required to calculate actual amounts of nitrogen fixation, 
using these estimates of %N from nitrogen fixation. Although it can be 
seen that estimation of nitrogen fixation is difficult, there is still remarkably 
little quantitative information given the potential importance of nitrogen 
fixation in the long-term sustainability of agroforestry systems. 

Chapter 14 


School of Agricultural and Forest Sciences, University of Wales, 
Bangor, Gwynedd, LL57 2UW, UK 

14.1 Synopsis 

Most trees and crop plants form mycorrhizal associations but our 
understanding of their role, particularly in mixed species associations, is 
still developing. Whereas trees form associations with both ectomycorrhizal 
(Fig. 14.1) and arbuscular mycorrhizal (or endomycorrhizal) fungi (see 
Fig. 15.4 on p. 295), herbaceous crop plants only form arbuscular 
mycorrhizas. There is strong evidence to show that mycorrhizal 
colonization can be of great benefit to plants but that effects of colonization 
vary markedly depending on environmental conditions and species of both 
host plant and fungus. The benefits to host plants are assumed to be 

Fig. 14.1. Pinus strobus root with ectomycorrhizal colonization. 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels G. Schroth and 
F.L. Sinclair) 


272 D.L. Codbold and R. Sharrock 

because mycorrhizal colonization may improve nutrient uptake, water 
relations, resistance to soil pathogens, and resistance more generally to 
adverse soil conditions. Furthermore, since mycorrhizas may infect many 
different host species, the mycelial network they develop below ground 
may connect plants and facilitate transfers of carbon and nutrients among 
them. In this chapter, the evidence for potential effects of mycorrhizal 
colonization is discussed, leading to the identification of key research 
questions and then description of the methods required to address them. 

Importance of mycorrhizas for plant nutrition 

The improved mineral nutrition of mycorrhizal plants is well documented 
(Marschner, 1995). In particular, a role in the uptake of phosphorus by 
ectomycorrhizas or arbuscular mycorrhizas, and nitrogen uptake by 
ectomycorrhizas and to a lesser extent arbuscular mycorrhizas, has been 
described (George et al., 1995). In Eucalyptus grandis and Eucalyptus 
maculata, inoculation with the ectomycorrhizal fungus Pisolithus sp. enabled 
the seedlings to utilize organic nitrogen sources, which could not be 
utilized by non-mycorrhizal seedlings (Turnbull et al., 1995). In addition 
to the elements phosphorus and nitrogen, mycorrhizas have been shown 
to facilitate plant acquisition of magnesium (Jentschke et al., 2000), copper 
and zinc (Guo et al, 1996) and manganese (Marschner, 1995). The effects 
of arbuscular mycorrhizas on mineral acquisition of plants have recently 
been reviewed by Clark and Zeto (2000). 

Both ectomycorrhizas and arbuscular mycorrhizas can greatly increase 
the volume of soil exploited due to the extent and high surface area of the 
extramatrical mycelium. It has been shown in a number of investigations 
that the external hyphae of mycorrhizas can absorb phosphorus from 
outside the root depletion zone and transport it to the host plant. Some 
of the most convincing demonstrations of this have used experimental 
designs with separate root and hyphal compartments. In these 
investigations roots were excluded from the hyphal compartment by a 30 
\\.n\ net which allowed mycorrhizal hyphae to pass. With such designs, in 
white clover (Trifolium repens) colonized by arbuscular mycorrhizas, a 
depletion zone of more than 1 1 cm due to phosphorus uptake by the 
extramatrical hyphae could be shown (Li et al, 1991). In an investigation 
with Norway spruce (Picea abies) and the ectomycorrhizal fungus Paxillus 
involutus, using a separate root and hyphal compartment, translocation of 
phosphorus by the extramatrical mycelium over a distance of 5 cm was 
demonstrated (Brandes et al., 1998). In Pinus sylvestris mycorrhizal with 
Suillus bovinus, phosphorus was translocated over 30 cm, mainly in 
rhizomorphs (Finlay and Read, 1986a). Effective uptake of phosphorus by 
the hyphae of both ectomycorrhizas and arbuscular mycorrhizas may also 

Mycorrhizas 273 

be because polyphosphates are accumulated in vacuoles, where they act 
as both a storage form of phosphorus and function in energy storage. 

In addition to the increase in surface area provided by the 
extramatrical mycelium, ectomycorrhizas have also been shown to exude 
organic acids and mobilize mineral phosphorus from sparingly soluble 
sources. The involvement of organic acid exudation in uptake of 
phosphorus in non-mycorrhizal roots is well documented (Marschner, 
1995). The ectomycorrhizal fungus Paxillus involutus exudes high amounts 
of malate and oxalate (Lapeyrie et al., 1987; Marschner, 1994). Gumming 
and Weinstein (1990) demonstrated that Pinus rigida ectomycorrhizal with 
Pisolithus tinctorius was able to extract phosphorus from insoluble 
aluminium phosphate. Recently, it has been suggested that 
ectomycorrhizas are a primary factor in the weathering of minerals 
(Jongmans et al., 1997; Landeweert et al., 2001). Thus there is a strong 
possibility that organic acids are involved in the uptake of phosphorus and 
also other nutrients such as potassium, calcium and magnesium 
(Landeweert et al, 2001) by ectomycorrhizas. 

Both ectomycorrhizas and arbuscular mycorrhizas have high levels of 
phosphatase activity. In wheat (Triticum aestivum), roots infected with 
arbuscular mycorrhiza had higher levels of acid phosphatase than non- 
mycorrhizal roots (Tarafdar and Marschner, 1994). In Norway spruce 
ectomycorrhizal with a number of fungi, similar levels of acid phosphatase 
were found in mycorrhizal and non-mycorrhizal roots (Eltrop, 1993). 
However, in Norway spruce mycorrhizal with Thelephora terrestris, a higher 
phosphomonoesterase activity was found in mycorrhizal roots and 
rhizomorphs than in non-mycorrhizal roots. These results indicate that 
mycorrhizas have a high potential to mobilize organic phosphorus, but 
that there are differences between species of mycorrhizas. However, even 
if the levels of acid phosphatase activity are similar between mycorrhizas 
and non-mycorrhizal roots, the large surface area of the extramatrical 
mycelium will greatly increase the potential to mobilize organic 

There are few investigations of the contribution of mycorrhizas to the 
total uptake of phosphorus by a plant. In Norway spruce ectomycorrhizal 
with Paxillus involutus, 52% of the total phosphorus uptake was shown to 
be via the extramatrical hyphae (Brandes et al., 1998). In arbuscular 
mycorrhizal Trifolium repens, the hyphae contributed 70-80% of the total 
phosphorus uptake (Li et al., 1991). In a similar system with Trifolium 
repens, phosphorus uptake in non-mycorrhizal plants was only 3% of the 
uptake of mycorrhizal plants (Schweiger et al., 1999). In non-mycorrhizal 
Leucaena leucocephala, soil solution phosphorus concentrations 27-38 times 
higher were required for maximal growth than in mycorrhizal seedlings 
(Habte and Manjunath, 1987). These data indicate that mycorrhizas have 
a high potential to contribute to phosphorus acquisition by plants. It is 

274 D.L. Codbold and R. Sharrock 

possible that experimental systems overestimate the mycorrhizal 
contribution to phosphorus acquisition but, since the mycorrhizal 
contribution measured experimentally is so high, it is likely that under 
field conditions mycorrhizas play a very significant role in phosphorus 

Importance of mycorrhizas for plant water relations 

Both ectomycorrhizas (Lamhamedi et al., 1992) and arbuscular 
mycorrhizas (Allen, 1991) can improve the water relations of plants. For 
example, in two species of Acacia, Cornet and Diem (1982) showed that 
mycorrhizal plants had a higher number of open stomata under drought 
conditions. However, the mechanisms involved are still controversial. A 
number of mechanisms of increased host drought tolerance have been 
suggested, such as increased root hydraulic conductivity (Koide, 1985), 
alteration of stomatal regulation due to hormone signals (Allen et al., 1982; 
Barea and Azcon-Aguilar, 1982), osmotic adjustment (Auge et al., 1986), 
hyphal water transport (Hardie, 1985; Faberetal., 1991; Ruiz Lozano and 
Azcon, 1995) and improved phosphorus nutrition (Fitter, 1988). 

Improved water status has often been associated with improved host 
phosphorus nutrition (Nelsen and Safir, 1982; Graham and Sylvertson, 
1984; Fitter, 1988). Mycorrhization nearly always results in an improved 
phosphorus nutrition and/or larger size of host plants, which often makes 
it difficult to deduce the mechanisms involved in improved water relations. 
However, a number of authors have shown that host water status is 
improved independently of host phosphorus nutrition (Auge et al., 1986; 
Bethlenfalvay et al, 1988; Peha et al., 1988; Sanchez-Diaz et al., 1990). 
Davies et al. (1992) increased the phosphorus nutrition of non-mycorrhizal 
plants by addition of fertilizers to produce pepper (Capsicum annum) plants 
mycorrhizal with Glomus deserticola and non-mycorrhizal pepper plants 
with similar levels of phosphorus. After drought acclimation, mycorrhizal 
plants were the most drought tolerant. These authors suggested that the 
extraradical hyphae might facilitate water uptake and improve drought 
resistance. Faber et al. (1991) have suggested a role of the hyphae in the 
transport of water. These authors suggested that hyphae of arbuscular 
mycorrhizas might have access to smaller pores within the soil and facilitate 
water uptake under drought conditions. 

To estimate the contribution of the extraradical hyphae to water 
uptake, techniques have been used with separate hyphal and root 
compartments, divided by a nylon screen. Using such techniques, George 
et al. (1992) found no evidence for the involvement of hyphae of arbuscular 
mycorrhizas in water transport of Agropyron repens. Using a similar 
experimental design, Ruiz Lozano and Azcon (1995) were able to show 

Mycorrhizas 275 

that, in association with lettuce (Lactuca sativa), extraradical hyphae of both 
Glomus deserticola and Glomus fasciculatum were involved in water uptake. 
However, differences in water uptake were shown for the two fungi under 
different water regimes. Under poor water supply Glomus deserticola 
transported a larger amount of water than Glomus fasciculatum, suggesting 
species differences in hyphal water uptake and transport. 

Effects of mycorrhizas on plant-pathogen interactions 

Apart from beneficial effects on the nutrition and water supply of host 
trees, mycorrhizas have been proven to increase the resistance of trees to 
infection by pathogens of the fine roots (Marx, 1972; Duchesne et al., 
1989). Different species of mycorrhizal fungi vary in their efficiency at 
preventing root infections, although the mechanisms involved in this 
process are rather poorly understood. Possible modes of action include 
the production of antibiotics by the mycorrhizal fungi, stimulation of host 
defence mechanisms and the physical barrier presented by the Hartig net 
in ectomycorrhizal roots (Marx, 1972). Gunjal and Patil (1992) showed 
that arbuscular mycorrhizal fungi increased the resistance of casuarina 
(Casuarina equisetifolia) to Fusarium vesicubesum. Infections of wheat with 
common root rot (Bipolaris sorokiniana) and root disease caused by 
Gaeumannomyces graminis and Fusarium culmorum were reduced by 
arbuscular mycorrhizal fungi (Thompson and Wildermuth, 1989; 
Innocenti and Branzanti, 1998). In the studies by Innocenti and Branzanti 
(1998), the colonization of the roots with Glomus did not reduce the disease 
index, but increased host tolerance to pathogen attack. Both 
ectomycorrhizal and arbuscular mycorrhizal fungi have been shown to 
produce antibiotics (Marx, 1969; Trofast and Wickberg, 1977). In addition, 
ectomycorrhizal fungi produce phenolic substances (Strobel and Sinclair, 
1991) and oxalic acid (Duchesne et al., 1989), which suppress root 
pathogens. Arbuscular mycorrhizal fungi have also been shown to reduce 
damage to plants by sedentary endoparasitic nematodes (Borowicz, 2001). 
An ectomycorrhizal fungus (Pisolithus sp.) increased the growth of 
nematode-sensitive Acacia holosericea seedlings under glasshouse conditions 
and reduced the multiplication of the root-knot nematode, Meloidogyne 
javanica, possibly through the production of polyphenolic compounds 
(Duponnoisrfa/., 2000). 

Transfer of carbon and nutrients between plants 

Ectomycorrhizas and arbuscular mycorrhizas form a large mycelia network 
in the soil. As trees of different species may be colonized by the same 

276 D.L. Codbold and R. Sharrock 

species of ectomycorrhiza, and arbuscular mycorrhizas mostly have a wide 
host range, different species of trees and other plants may be connected 
by a common below-ground hyphal network. It has often been suggested 
that this hyphal network of mycorrhizas forms a pathway for the 
movement of carbon and nutrients between plants (Finlay and Read, 
1986b). In forests with ectomycorrhizal trees a movement of carbon 
between trees has been shown. In laboratory studies, transfer of carbon 
has also been demonstrated between plants connected by an arbuscular 
mycorrhizal mycelium (Francis and Read, 1984). 

Although the transfer of carbon between species or between mature 
trees and seedlings of a species is of great ecological interest, in 
agroforestry and agricultural systems the transfer of nutrients, especially 
nitrogen and phosphorus, is more important. This is particularly the case 
for the transfer of nitrogen from nitrogen-fixing trees or crops to receiver 
plants. This invariably requires the movement of nutrients between 
arbuscular mycorrhizal plants. There are conflicting reports about the role 
of mycorrhizas in the transfer of nitrogen between plants (Hamel et al., 
1991; Frey and Schiiepp, 1993). Frey and Schiiepp (1993) showed that the 
mycorrhizal mycelium enhanced the transfer of nitrogen between clover 
(Trifolium alexandrinum) and maize (Zea mays). In contrast, Hamel et al. 
(1991) found that the transfer between soybean (Glycine max) and maize 
was small. The transfer increased if there was direct contact between the 
roots. Between soybean and maize a greater transfer occurred for nitrogen 
supplied from fertilizer than nitrogen from biological fixation (Marschner, 
1995). This suggests that plants restrict the movement of symbiotically 
fixed nitrogen. There is little evidence to show that there is a transfer of 
nitrogen from the plant to the fungus (Ikram et al., 1994). 

The direct transfer of nitrogen between plants may, however, be a 
function of the sink-source relationship between potential donor and 
receiver plants (Frey and Schiiepp, 1993). Ekblad and Huss-Danell (1995) 
showed that in a mixture oiAlnus incana and Pinus sylvestris the proportion 
of fixed nitrogen in P. sylvestris derived from A. incana was highest (9%) 
when P. sylvestris was nitrogen starved. Ikram et al. (1994) investigated 
nitrogen movement between Pueraria phaseoloides as a nitrogen-fixing 
donor plant and rubber trees (Hevea brasiliensis) as receiver plants. 
P. phaseoloides is used as a cover crop in rubber plantations. In the 
investigation, the shoot of the donor plant was shaded or removed to 
increase the difference in sink strength in favour of the receiver plant. 
Despite this treatment P. phaseoloides only supplied 0.07-0.4% of the total 
nitrogen uptake ofH. brasiliensis. This suggests that mycorrhizal links play 
only a small role in the transfer of nitrogen in this system. However, it is 
known that intercropping of rubber with nitrogen-fixing P. phaseoloides, as 
in other intercropping systems, increases the growth of rubber 
(Broughton, 1977). The roots and mycorrhizas may take up nitrogen 

Mycorrhizas 277 

released by the decomposition of leaf and root litter. It has also been 
suggested that nitrogen may be transferred via root exudation of nitrogen 
compounds (see Table 13.1 on p. 262). Transfer from the roots may also 
be higher if the roots of the donor plants are dying. Johansen and Jensen 
(1996) found a significant transfer of nitrogen from pea (Pisum sativum) to 
barley (Hordeum vulgare) in mycorrhizal systems when, after removal of the 
shoot, the roots of the donor plant were dying. Mycorrhizas are able to 
take up and transport both inorganic and organic forms of nitrogen 
(George et at., 1995), both of which may be available in dying roots. Such 
a transfer may be of particular significance after pruning of nitrogen-fixing 
trees in agroforestry systems, and in trees with high rates of root turnover. 

Differences between mycorrhizal species and strains 

In an investigation of the distribution of arbuscular and ectomycorrhizal 
tree species in savannas, Hogberg (1989) suggested that ectomycorrhizal 
trees dominate soils with low levels of nitrogen and phosphorus, whereas 
arbuscular mycorrhizal species dominate soil low in phosphorus but 
relatively rich in nitrogen. This suggests that ectomycorrhizal trees may 
have a greater capacity for nitrogen acquisition than arbuscular 
mycorrhizal species. However, in the humid tropics of French Guiana, 
ectomycorrhizal tree species were found not to dominate the poorest soils 
(Rereauetai, 1997). 

The considerable benefits of mycorrhization can be seen in the 
advantages conferred by efficient species and strains. There are, however, 
significant differences between the effects that different species and 
strains of both ectomycorrhizas and arbuscular mycorrhizas have on host 
plants. It has been shown that the potential to take up phosphorus varies 
strongly between species both for ectomycorrhizas and arbuscular 
mycorrhizas (Marschner, 1995). As described above, species differences 
have also been demonstrated for improved water relations and pathogen 
resistance. Similarly, there are considerable functional differences between 
species of mycorrhizas and different species of trees. Ba et al. (2000) 
demonstrated that only five species of tropical fruit trees (Zizyphus 
mauritiana, Tamarindus indica, Dialium guineensis, Parkia biglobosa and Cordyla 
pinnata) out of a total of 15 species had higher dry weight as a result of 
inoculation with Glomus aggregatum. In Zizyphus mauritiana and Cordyla 
pinnata a smaller positive effect was shown after inoculation with Glomus 

At present there is a potential danger of either underestimating or 
overestimating the significance of mycorrhizas because only a small 
number of mycorrhizal species have been investigated. For example, to 
reach reliable estimates of the contribution of mycorrhizas to phosphorus 

278 D.L. Codbold and R. Sharrock 

uptake by plants in the field, we need to know the species composition of 
the mycorrhizas colonizing roots and have information about the 
phosphorus uptake capabilities of each species. The development of 
molecular biological methods of species identification for both 
ectomycorrhizas and arbuscular mycorrhizas has greatly increased our 
understanding of species composition (see Section 14.5). 

Inoculation and management of mycorrhiza populations 

Considering the advantages conferred by mycorrhizal infection, it is easy 
to understand why inoculation with mycorrhizas normally greatly increases 
growth and survival of host plants over that of non-mycorrhizal plants. 
Increased growth as a result of inoculation with mycorrhiza has been shown 
on numerous occasions in plants used in agroforestry systems (Osonubi et 
a!,., 1991). However, the majority of studies that have investigated the 
advantages of mycorrhizal infection have compared mycorrhizal plants with 
non-mycorrhizal plants, which is unrealisitic since in most soils, with the 
exception of soils with an extremely low inoculation potential, most plants 
will form some sort of mycorrhizal association albeit weak. 

However, with an increasing knowledge of the importance of 
mycorrhizal species differences, there is an increasing interest in the 
introduction of mycorrhizal species that are more efficient than the species 
present in a given soil under field conditions. Once introduced, these 
mycorrhizas must persist in the soil together with the indigenous 
mycorrhizas. Due to the difficulties in species identification of mycorrhizas, 
especially arbuscular mycorrhizas, there are few studies to estimate the 
persistence of introduced mycorrhizal species. With the development of 
better identification methods a clearer picture of mycorrhizal persistence 
will emerge. It is reasonable to assume that the highest success rate for 
introduced mycorrhizas will be achieved in soils with the least competitive 
indigenous mycorrhizal flora. However, even if introduced mycorrhizal 
species have a low persistence, the advantages conferred on the plant 
during establishment may be sufficient to justify their use. This may be 
particularly important for trees under adverse conditions, where 
establishment is often the biggest problem. 

Management options for mycorrhizal populations in agricultural 
systems have been reviewed several times (Barea and Jeffries, 1995; 
Haselwandter and Bowen, 1996). Briefly these include using crop rotation 
techniques, precropping or intercropping. In each case a plant species is 
used to maintain a high inoculum potential in the soil. For example, a high 
level of mycorrhizal infection was maintained in cassava (Manihot esculenta) 
when it was grown in rotation with groundnut (Arachis hypogaea) 
(Sieverding and Leihner, 1984). Dodd et al. (1990a, b) showed that 

Mycorrhizas 279 

precropping with cassava, kudzu (Pueraria phaseoloides) or sorghum 
(Sorghum bicolor) significantly increased arbuscular mycorrhizal infection 
of cowpea (Vigna ungukulata). As intercrops, millet (Pennisetum glaucum), 
Sudan grass (Sorghum sudanense) and sorghum increased the population 
densities of arbuscular mycorrhizal fungi and improved the growth of tree 
seedlings (Johnson and Pfleger, 1992). The reverse may also be true 
where, in agroforestry systems, highly mycorrhizal trees may be used to 
maintain the inoculum potential in the soil. Roots of Citrus aurantifolia 
grown in low-input systems had a higher degree of mycorrhizal 
colonization than those grown in conventional high-input systems in 
Mexico (Michel-Rosales and V aides, 1996). These effects may have been 
related to fertilizer input, but the low-input systems were also structurally 
diverse, with a large number of other species including 16 species of fruit 
trees, compared with the high-input systems, which were monocultures. 
However, on a Nigerian Alfisol, Atayese et al. (1993) found that Senna sp., 
Gliricidia sp. and Leucaena sp. did not increase the mycorrhizal spore 
density of the soil even though Gliricidia sp. and Leucaena sp. were highly 
mycorrhizal. The mycorrhizal spore density was, however, increased by 
planting cassava (Manihot esculenta). 

Research needs 

Many of the studies cited above have been carried out under laboratory 
conditions. Therefore, there is a general requirement to transfer results 
obtained in laboratory experiments to the field. Many of the questions that 
need to be answered for mycorrhizas in tree-crop associations, listed 
below, are similar to the questions relevant for conventional agriculture 
and forestry. 

• Considerably more information is required about the ecophysiology 
and effectiveness of different mycorrhizal species, as well as their 
below-ground population dynamics. Information about species should 
include estimates of which nutrient pools mycorrhizas are accessing 
and whether these differ from non -mycorrhizal plants. 

• Information is also required about the persistence of introduced 
mycorrhizas, and whether persistence can be increased by cultivation 
techniques. Additionally, information is required on how management 
actions, such as mulching, crop rotation and fertilization (both mineral 
and organic), affect the development of mycorrhizas. 

In addition there are some research questions specific to tree-crop 

D.L. Codbold and R. Shar 

Can the association of crops with strongly mycorrhizal trees or tree 
species with particularly efficient mycorrhizas increase the access of 
these crops to less available nutrient pools in the soil via nutrient 
release from tree litter? 

Can strongly mycorrhizal species aid the establishment of poorly 
mycorrhizal species under adverse site conditions, for example in 
rehabilition of degraded lands? 

14.2 Inoculation Methods 

Propagules of mycorrhizal fungi include soil hyphae, spores, old root 
fragments and colonized particles of organic matter (Brundrett and 
Abbott, 1994). In particular, soil hyphae are important propagules under 
drought conditions. Soil disturbance and degradation often result in low 
inoculation potential of the soil; hence there is often a need to reintroduce 
mycorrhizal fungi in general or to introduce efficient mycorrhizal strains 
or species. Land management practices can then be used to maintain 
introduced mycorrhizas or enhance indigenous populations. 

The simplest source of mycorrhizal propagules is soil from areas of 
high mycorrhizal activity. Soil may be used as a non-specific inoculum for 
both ectomycorrhizal and arbuscular mycorrhizal fungi. However, for 
reasons of phytosanitation it is often undesirable to use soil as inoculum. 
For this reason and also to allow introduction of specific fungi, artificial 
inocula are often required. 

For ectomycorrhizas, inoculation techniques have been developed 
which use spores of selected fungi (Marx et a!,., 1984). Mycelium of 
ectomycorrhizal fungi can easily be produced in fermenters. The mycelium 
can then be trapped in alginate beads for use as inoculum (Mauperin et 
al., 1987). 

Arbuscular mycorrhizal fungi cannot be cultured ex planta and thus 
must be cultured with a host plant. Most inoculum production requires 
growth of the host plants in a medium that can be used as a carrier for 
spores and hyphae. Carrier substances include expanded clays (Feldmann 
and Idczak, 1994) or vermiculite or perlite (Bagyaraj, 1994). Soil-free 
inoculum has been produced in aeroponic (Jarstfer and Sylvia, 1995) and 
nutrient-film cultures (Elmes and Mosse, 1984). The suitability of these 
inocula for different applications depends on the identity of the main 
mycorrhizal propagules and their ability to retain infectivity during storage 
and to persist in the soil from year to year, as well as on methods available 
for application (Smith and Read, 1997). Due to the expense of the 
production of inoculum and the bulk required, large-scale inoculation is 
often not a viable option. Thus management of indigenous populations is 
of great importance (see Section 14.1). 

Mycorrhizas 281 

14.3 Sampling of Plant Roots for Mycorrhizal Studies 

Collection of plant roots for estimating mycorrhizas requires sampling, 
preparation and storage of the roots. Root samples are normally collected 
using a trowel or soil auger (see Section 12.2). Roots are normally removed 
from the soil using careful washing with water or removal of the soil using 
forceps. Particularly for visual identification of ectomycorrhizas damage 
to the roots should be avoided. For quantification and identification (both 
visual and molecular) of arbuscular mycorrhizas and for molecular 
identification of ectomycorrhizas, roots can be stored in 50% (v/v) ethanol. 
For visual quantification and identification of ectomycorrhizas, roots can 
be stored in moistened Petri dishes at 4°C for several days. Fixation of 
ectomycorrhizas often results in colour changes, making visual 
identification more difficult. 

14.4 Quantification of Mycorrhizas 

Non-vital staining of arbuscular mycorrhizas 

In order to visualize and quantify root colonization by arbuscular mycor- 
rhizal fungi the fungal structures must be stained. The most commonly 
utilized stains are non-vital stains, which cannot distinguish between living 
and dead fungal tissues. Stains for light microscopy include trypan blue 
(Koske and Gemma, 1989) and chlorazol black E (Brundrett et at., 1984). 
Acid fuchsin is also a popular non-vital stain; this stain has fluorescent 
properties, which are best visualized by fluorescence microscopy 
(Merryweather and Fitter, 1991). 

One of the most frequently cited non-vital staining procedures 
utilizes the fungal staining properties of trypan blue (Phillips and 
Hayman, 1970; Koske and Gemma, 1989). Roots are washed from the 
soil and either stained directly or fixed in 50% ethanol and stained 
at a later time. The root tissue is cleared in 2.5% KOH at 90 °C for 
10-30 min in a water bath. After clearing, dark tissues may need 
to be bleached in a freshly prepared solution of alkaline hydrogen 
peroxide for 10-45 min. Following the KOH and bleaching steps, the 
root tissue is highly alkaline and requires acidification prior to staining 
with trypan blue. The roots are rinsed thoroughly and acidified in 
1% HC1 for 1-24 h (at least 8 h is generally recommended). The acidified 
roots can then be stained in 0.05% trypan blue in acidic glycerol at 90 °C 
for 15-60 min. The stained tissues are then de-stained in acidic glycerol 
at room temperature. De-staining overnight generally improves the 
contrast between the fungal and root tissues, which often initially retain 
excess stain. Stained tissues can be stored in acidic glycerol in the dark for 

282 D.L. Codbold and R. Sharrock 

many months. For the protocol details in full see Koske and Gemma 

Visual methods of evaluation of arbuscular mycorrhizas 

The assessment of fungal colonization is generally based on estimates of 
the proportion of potential host tissue (the root cortex) colonized by 
arbuscular mycorrhizal fungi (Giovannetti and Mosse, 1980; McGonigle 
et at., 1990). The most extensively used visual method of determining 
fungal colonization is the grid-line intersect method (Giovannetti and 
Mosse, 1980; McGonigle et at., 1990). This method is used to estimate the 
proportion of colonized tissue and total root length. A root sample is 
spread across a grid of known dimensions drawn on the base of a Petri 
dish. Using a dissecting microscope at x40 magnification, each point at 
which a root intersects a vertical or horizontal grid line is studied and the 
presence or absence of fungal tissue is noted. 

However, at x40 magnification it is difficult to distinguish between 
different fungal structures. In light of this McGonigle et at. (1990) devised 
an objective method of measuring root colonization called the magnified 
intersection method. This method is used to obtain the percentage 
colonization of different fungal structures in a root system. Root sections, 
approximately 1 cm in length, are mounted on microscope slides in a 
few drops of acidic glycerol and viewed with a compound microscope 
at magnification x200. Root/eye-piece cross-hair intersections are 
considered along the axis of the root section and the presence of 
arbuscules, hyphae and vesicles is noted at each intersection. A fixed 
number of intersections are considered per root section. From these data 
the percentage arbuscular, hyphal and vesicular colonization can be 

Visual methods of evaluation of ectomycorhizas 

The degree of colonization of ectomycorrhizas can be estimated by 
counting the number of mycorrhizal and non-mycorrhizal root tips in a 
representative sample (> 100 root tips). 

Vital staining 

To determine the proportion of active fungal colonization associated with 
a root, vital staining procedures are employed. Vital stains locate sites of 
enzymatic activity, which are specific to viable fungal structures. Nitroblue 

Mycorrhizas 283 

tetrazolium is a vital stain which is coupled to the activity of succinate 
dehydrogenase (a tricarboxylic acid enzyme), to produce formazan, which 
is purple (Smith and Dickson, 1991; Schaffer and Peterson, 1993). Used 
in conjunction with a non-vital stain such as acid fuchsin, the proportion 
of active to total colonization can be assessed. 

Alkaline phosphatase activity is also considered to be a good indicator 
of fungal viability. Alkaline phosphatase is sequestered in the phosphatase- 
accumulating vacuoles in hyphae and is also found along the fungal 
tonoplast, but is not present in root tissue (Dickson and Smith, 1998). 
Other vital (and non-vital) stains are reviewed by Dickson and Smith 

Laser scanning confocal microscopy 

Visualization of mycorrhizal structures and quantification of their surface 
area and volume have recently been achieved by laser scanning confocal 
microscopy after staining with acid fuchsin (Dickson and Kolesik, 1999). 

Chitin assay 

The amount of fungal tissue present in the host tissue can be measured by 
chemical means. The chitin assay is based on the quantitative 
determination of chitin, which is present in the fungus, but not in the host. 
Total chitin is measured by colorimetric assay after conversion into 
glucosamine (Hepper, 1977; Bethlenfalvay et ai, 1981). 

Molecular quantification 

Quantification of endomycorrhizal fungi in a root system has also been 
achieved by molecular means using an assay based on competitive 
polymerase chain reaction (PCR) (Edwards et a!,., 1997). 

14.5 Identification of Mycorrhizal Fungi 

Visual identification 

Many ectomycorrhizas can be identified using a visual method developed 
and described by Agerer (1992, 1993). This method involves using both 
morphological features and the hyphal mantel characteristics to identify 
to genus or species level. Visual methods can also be used to identify some 

284 D.L. Codbold and R. Sharrock 

arbuscular mycorrhizas; however, this method is of limited potential as a 
single mycorrhizal species may have a different morphology between host 
plant species. 

Immunological approaches 

Immunochemical detection methods are based on the reaction between 
antibodies and antigens. The production of antibodies is an immune 
system response to alien substances (the antigen). The type of antibody 
produced is often antigen-specific. It is this specific relationship between 
antibody and antigen that can be utilized as a tool in fungal identification 
(Gobel et al., 1998; Hahn et al, 1998). Antibodies, produced by exposing 
an animal's immune system to a specific antigen, are coupled by chemical 
means to molecules which can be easily detected, such as fluorescent dyes 
for fluorescence microscopy, enzymes for enzyme-linked immunoassays, 
or heavy metals for immunocytochemical analysis with an electron 
microscope (Harlow and Lane, 1988; Gobeletal., 1998; Hahnetal., 1998). 
Labelled antibodies introduced into a system in which the fungal symbiont 
is unknown will recognize and bind to their corresponding antigen 
(providing it is present). Schmidt et al. (1974) were among the first to 
analyse mycorrhizal fungi by immunochemical means. More recently, 
polyclonal antibodies have been generated with sufficient stringency to 
detect Scutellospora species in the roots of several plants (Thingstrup et al, 
1995). Dot immunoblots have also been used in an attempt to identify 
Gigaspora and Acaulospora species (Sanders et al., 1992). 

Isozyme analysis 

Isozymes are different molecular forms of the same enzyme. They 
generally have the same enzymatic properties and differ only in their 
amino acid composition and net charge. Due to these differences in net 
charge, isozymes can be differentiated by their electrophoretic mobility 
(Rosendahl and Sen, 1994). Many isozymes are apparently species-specific 
in nature. Tisserant et al. (1998) ran protein extracts taken from root tissue 
colonized by five Glomus species on a polyacrylamide gel. The separated 
protein bands were stained for ten different enzymes, including alkaline 
phosphatase and malate dehydrogenase. Different banding patterns 
emerged, revealing the presence of several mycorrhizal-specific isozymes. 
These mycorrhizal-specific isozymes were subsequently successfully used 
as fungal-specific markers in plants grown in field trials. 

Isozyme analysis is also increasingly used in conjunction with PCR- 
and restriction fragment length polymorphism (RFLP)-based molecular 

Mycorrhizas 285 

approaches to identifying both endomycorrhizal (Dodd et a!,., 1996) and 
ectomycorrhizal (Timonen et al., 1997) fungi in root tissue. 

Mycorrhizal identification by molecular means 

Molecular approaches to mycorrhizal identification are becoming 
increasingly popular. Wyss and Bonfante (1993) extracted genomic DNA 
from the spores of Glomus versiforme and Gigaspora margarita isolates. 
Genomic fingerprints of these endomycorrhizal isolates were generated 
by PCR amplification with short arbitrary primers, a process known as 
random amplified polymorphic DNA (RAPD-PCR). The RAPD-PCR 
method employs an arbitrary 10 base-pair primer, which anneals to a 
number of complementary sites on the template DNA. The PCR products 
generated are separated according to their molecular weights on an 
agarose gel and stained with ethidium bromide. The banding pattern on 
the gel reflects the overall structure of the DNA molecule used as the 
template. If the starting material is total cell DNA, then the banding 
pattern represents the organization of the genome (Williams et al., 1990). 
Wyss and Bonfante (1993) found that different mycorrhizal species 
generated different banding patterns. To a lesser extent, differences were 
also observed between different isolates of the same species. 

Isolate-specific RAPD-PCR bands have been used to generate isolate- 
specific PCR primers by Abbas et al. (1996). The DNA present in bands 
unique to isolates of Glomus mosseae and Gigaspora margarita was purified, 
cloned and sequenced. Sequence analysis subsequently led to the 
generation of isolate-specific primer pairs, which in conjunction with PCR 
technology was utilized to identify these isolates in root systems. The 
presence of a particular isolate in a root system is identified by the presence 
of a PCR product band on an agarose gel. 

Ribosomal DNA (rDNA) sequences have also been extensively utilized 
in the generation of mycorrhizal-specific PCR primers. Using PCR, RFLP, 
cloning and sequencing technologies, sequence differences have been 
highlighted in the following rDNA regions: the small subunit (SSU)/18S 
gene, the large subunit (LSU)/28S gene, internal transcribed spacer (ITS) 
regions and intergenic spacer (IGS) regions (see Egger, 1995). These 
differences are apparently family/species/isolate specific and have been 
used to generate mycorrhiza-specific PCR primers. 

Simon et al. (1993) used single subunit (SSU)/18S gene sequences from 
a number of endomycorrhizal fungi to generate family-specific primers. 
The SSU sequences were aligned and regions apparently unique to a 
family were selected for the design of family-specific primers. Members of 
the Acaulosporaceae, Gigasporaceae and Glomaceae have been successfully 
identified using these family-specific primers in spores and root systems. 

D.L. Codbold and R. Shar 

Van Tuinen et al. (1998) utilized the LSU/28S rDNA region to generate 
species/isolate-specific primers for the endomycorrhizal fungi Glomus 
mosseae, Glomus intraradices, Scutellospora castanea and Gigaspora rosea. 
Comparable studies have been carried out on endomycorrhizas by Sanders 
et al. (1995), Clapp et al. (1999) and Helgason et al. (1999). 

Ectomycorrhizas have also been subject to molecular analysis. Erland 
(1995) studied the abundance of Tylospora fibrillosa ectomycorrhizas 
in a spruce forest in Sweden. Tylospora fibrillosa could be distinguished 
from a large number of other basidiomycetes by the banding pattern 
generated by PCR amplification of the rDNA ITS region followed by RFLP 
analysis of the PCR product. Paolocci et al. (1999) also utilized the rDNA 
ITS region to identify Tuber species, which were successfully characterized 
on host plants by PCR amplification using species-specific ITS primers. 
The analysis of rDNA has also been used to compare the ericoid 
mycorrhizas Scytalidium vaccinii and Hymenoscyphus ericae (Egger and Sigler, 

Indirect methods of quantification using spores and sporocarps 

The population structure of arbuscular and ectomycorrhizas can 
be estimated using soil spores for arbuscular mycorrhizas (Tommerup, 
1994), or sporocarps (mushrooms) for ectomycorrhizas. The spores of 
arbuscular mycorrhizas must be isolated from the soil (Pacioni, 1994) and 
can be identified to genus or species level using morphological 
characteristics (Brundrett et al., 1996). For ectomycorrhizas, sporocarps 
within an area can be collected, identified and their abundance estimated. 
However, the spore and sporocarp production is dependent upon a 
number of seasonal factors (temperature, moisture), and often correlates 
poorly with the below-ground abundance (Tommerup, 1994; Erland, 
1995). Additionally, a number of important ectomycorrhizas do not form 

14.6 Estimation of Mineral Nutrient Uptake Through 

Absorption, translocation and transfer of mineral nutrients by the external 
hyphae of mycorrhizas have been demonstrated a number of times using 
radioisotopes (Tinker et al., 1994). Most of the attempts at quantification 
of mineral nutrient uptake by mycorrhizal fungi have used techniques in 
which separate root and hyphal compartments have been established using 
fine mesh (Li et al., 1991; George et al., 1995; Schweiger et al., 1999; 
Jentschke et al, 2000). The size of the mesh is usually between 20 and 45 



|J,m, which allows mycorrhizal hyphae to penetrate, but restricts growth of 
roots. The proportion of mineral nutrients taken up by the mycorrhizas 
is normally estimated by determining the decrease in mineral elements in 
the soil (Li et at, 1991), or by using radioisotopes (Schweiger et at, 1999) 
or stable isotopes (Jentschke et at, 2000) as tracers. 

Transfer of nitrogen from nitrogen-fixing plants 

Techniques to measure the transfer of nitrogen from nitrogen-fixing 
plants to non-nitrogen-fixing plants via mycorrhizal hyphae have used 
compartments to prevent root contact as described above (Ikram et at, 
1994; Ekblad and Huss-Danell, 1995). A mesh is used to allow only 
mycorrhizal hyphae of the donor plant to be in contact with roots or 
hyphae of the receiver plants (Fig. 14.2). In some cases legumes grown in 
split root cultures have been used (Reeves, 1992). Nitrogen transfer is 
estimated using 16 N as a tracer. 15 N4abelled fertilizers are applied to the 
donor plants, and the amounts of fixed and transferred nitrogen is 
calculated using 15 N-dilution methods (Giller et at, 1991). 15 N is 
determined by mass spectrometry. 

20-45 |jm mesh 


Fig. 14.2. Experimental set-up used to investigate transfer of nitrogen from 
nitrogen-fixing plants to non-nitrogen-fixing plants via mycorrhizal hyphae. Root 
and hyphal soil compartments are separated by a fine mesh of pore size 20-45 
um, which allows passage of hyphae but not roots. 

Chapter 15 
Rhizosphere Processes 

D. Jones 

School of Agricultural and Forest Sciences, University of Wales, 
Bangor, Gwynedd LL57 2UW, UK 

15.1 Synopsis 

The rhizosphere can be defined as a zone of intense biological and 
chemical activity in the soil that surrounds the root. The rhizosphere is 
chemically, physically and biologically different from the bulk soil because 
it has been modified by roots and their associated microorganisms. Many 
key processes associated with soil fertility occur in this modified soil and 
the rhizopheres of different plants may interact. Where there are very high 
root densities, as in silvopastoral systems, most of the top 15-20 cm of soil 
may be modified by roots in this way, so that it all effectively constitutes 
rhizosphere soil. A comprehension of the rhizosphere is fundamental to 
understanding the interactions of plants with their environment and their 
ability to respond to stress. Different plants modify the environment 
around their roots in different ways and, therefore, interactions between 
plants may occur as a result of rhizosphere processes. The key rhizosphere 
processes that affect interactions between trees and soil fertility can be 
classified under three headings: 

• chemical processes, especially those affecting nutrient capture (see 
Chapter 8) and metal toxicity (see Chapter 5); 

• biological processes, including symbiotic associations (see Chapters 
13 and 14), pathogenic activity and allelopathy (where the release of 
chemicals from roots of one plant affect other plants); and 

• physical processes, including modification of soil structure (see 
Chapter 10) and its impact upon water availability (see Chapter 11). 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 289 

F.L. Sinclair) 

290 D.Jones 

The vast majority of research that is done on plant interactions with 
soil in agroforestry focuses on gross effects of, and changes in, soil 
properties, rather than the detailed rhizosphere mechanisms through 
which these come about, but it is necessary to understand basic processes 
if results are to be generalized and extrapolated. Much rhizosphere 
research has concentrated on only a few model plant species, in very 
controlled conditions, and its relevance to tropical agroforestry is unclear. 
This means that there is considerable scope to extend rhizosphere research 
to determine underlying mechanisms governing interactions in 
economically important tree and crop associations that are used in 

Potential importance of rhizosphere processes in agroforestry 

Agroforestry systems employ two or more, often contrasting, plant species 
such as shade trees and herbaceous grasses in silvopastoral systems, or 
cereal crops and nitrogen-fixing trees in hedgerow intercropping. The 
premise of these systems is that the interactions between associated plants 
with respect to below-ground resources such as nutrients and water are 
complementary or facilitative rather than competitive (see Section 5. 1). In 
some cases, one of the species may be partly chosen to stimulate above- 
and below-ground nutrient cycling (typically of phosphorus and nitrogen), 
as in the case of leguminous cover crops, legume shade trees and planted 
fallows. The quantification of this enhanced cycling can be achieved at a 
gross scale using stable isotopes and radiolabelled tracers ( 13 C, 15 N, 32 P, 33 P, 
35 S; see Sections 7.3 and 8.2). However, this provides little mechanistic 
information. In some cases it can be shown that co-planted species have 
different rooting strategies that result in nutrients being taken up from 
different spatial zones in the soil, but in other cases rooting patterns do 
not provide a mechanistic basis for the complementary or competitive 
effects observed and the mechanisms involved can only be speculated upon 
(Kamara etal, 1999; Lehmann et al. , 1999a; Ndufada/., 1999). 

To fully understand why these effects occur requires an understanding 
of nutrient flows in the rhizosphere. For example, where enhanced 
phosphorus cycling is observed, the impact of both mycorrhizal and root 
phosphorus mobilization strategies needs to be compared in each species 
to identify the spatial and temporal mobilization of different soil 
phosphorus pools such as organic phosphorus, iron-aluminium 
hydroxide-fixed phosphorus and calcium-bound phosphorus (see Section 
5.3). Furthermore, the plant species chosen for agroforestry systems may 
form different mycorrhizal associations (see Chapter 14). Trees may be 
either endo- or ectomycorrhizal or both, such as many species of the 
Eucalyptus genus, whereas associated crops may only be endomycorrhizal, 

Rhizosphere Processes 291 

as is the case with maize, or may not form mycorrhizal associations at all, 
such as Brassica spp. It can be envisaged that under these circumstances 
the mycorrhizas will develop vastly different rhizospheres leading to 
contrasting patterns of nutrient capture, although experimental evidence 
for this form of complementary resource use is lacking. In situations where 
the woody plant has the same mycorrhizal association as the crop, such as 
a Tithonia diversifolia hedge next to a maize crop, then competition for soil 
resources maybe more intense. However, as the rhizosphere of each plant 
species can be assumed to be unique, it is likely that resource capture by 
plants will always differ to some extent. 

The presence of one root system, such as that of a tree, can also aid in 
the resource capture by a crop plant. This is evidenced by both the direct 
and indirect transfer of nitrogen from nitrogen-fixing plants to non- 
nitrogen-fixing plants (see Chapter 13). In addition, a number of other 
interactions may occur where trees with abundant mycorrhizas can release 
root exudates capable of mobilizing the large reserves of organic nitrogen 
or phosphorus, making it available for associated crops. While our 
knowledge of the rhizosphere characteristics of different plant species 
remains so sparse we can only speculate upon the many interactions that 
may occur in species associations. 

Physical attributes of soil are also modified by roots in the rhizosphere 
and these strongly influence the growth and activity of organisms living 
in soil. Root exudates have been shown to significantly improve soil 
structure (see Chapter 10). This can occur directly in response to the root 
addition of polysaccharide gel (mucilage) to the rhizosphere, which helps 
bind aggregates together and also enhances soil-root contact (Czarnes et 
al., 2000). Indirectly, the release of both high and low molecular weight 
root exudates can also enhance soil structural stability through the 
stimulation of the rhizosphere microbial community, which in turn 
produces polysaccharide gels (Traore et al., 2000). As plants release 
different amounts of root exudates it can be expected that they influence 
soil physical properties in different ways. This is of relevance in an 
agroforestry context, where the maintenance or rehabilitation of soil 
structure is required and so plants may be selected specifically for their 
ability to enhance soil structure. 

Exudation of moisture by roots into dry rhizosphere soil can also be 
significant in wetting soil and so facilitating nutrient uptake by the plant 
whose roots exude the water as well as nutrient and water uptake by other 
plants in the vicinity (see discussion of hydraulic lift in Section 11.1). This 
clearly has implications for nutrient capture from otherwise dry soil and 
both competitive and facilitative interactions among species with different 
rooting habits. 


D. Jones 

Conceptual and methodological difficulties of rhizosphere studies 

Until recently, it has been difficult to achieve a mechanistic understanding 
of below-ground interactions between plants and the importance of 
rhizosphere processes in these interactions because techniques have not 
been available for unravelling the complexities of the processes involved. 
Indeed, we are still far from gaining a holistic understanding of the 
rhizosphere as we are only able to gain insights into specific areas for which 
techniques are available. Although there is conclusive evidence that the 
rhizosphere exists, its spatial boundaries remain difficult to define. This 
means that when embarking on rhizosphere research, a key step is to 
define an effective rhizosphere for the purposes at hand. In the case of 
chemicals lost from the root it is known that some may have an extensive 
diffusion sphere of many centimetres, whereas other highly charged 
compounds may only travel a few micrometres away from the root 
(Darrah, 1991a). Although they differ according to the chemical involved, 
the vast majority of chemical processes in the rhizosphere occur within a 
few millimetres of the root (Fig. 15.1). The exact definition of the 
rhizosphere and consequently the volume of soil that needs to be studied 
has to be relevant to the hypotheses being tested. For example, when 
studying allelopathic interactions in the rhizosphere, the effective radius 
of the rhizosphere is controlled by the diffusion of the allelopathic 

Mineral nutrients 

Organic carbon 

1.00 1.25 1.50 1.75 

Distance from root surface (mm) 

1.25 1.50 1.75 

Distance from rootsurface (mm) 

Fig. 15.1. Mineral nutrient and carbon profiles in a rhizosphere after 10 days 
according to simulation models based on experimentally derived nutrient uptake 
rates of maize and wheat. The mineral nutrient concentrations are relative to the 
amount present in the bulk soil at day (adapted from Barber, 1995, and Darrah, 

Rhizosphere Processes 


Epidermal cell 
Root cap mucilage mucilage 

Bacterial cell 

Epidermal cell death 
and loss 

Cortical cell death 
and loss 

Root cap 

Wound site 
(e.g. soil particle abrasion damage, 
where a secondary root emerges) 

Fig. 15.2. Different functional zones of a root; root hairs are not shown. 

Leaving the boundaries aside, there is also a high degree of both 
spatial and temporal complexity. Questions therefore arise as to whether 
roots should be treated as whole units or subdivided based on size, age or 
functional zone (Fig. 15.2). For example, root hairs (not shown in Fig. 15.2) 
can be usefully distinguished from the root apex. It is certainly clear from 
a root physiology point of view that the behaviour of different root areas 
is very different (Marschner, 1995). These functional root units can arise 
because of localized suberization/lignihcation, the degree of apoptosis or 
other forms of cell death, the different water and chemical status of roots 
in different soil environments or the degree of mycorrhizal infection. It is 
for this reason that most rhizosphere experiments are conducted in 
laboratory mesocosms under optimal growth conditions rather than in the 
field, where spatial heterogeneity makes it difficult to interpret residts. 

Typically, laboratory rhizosphere experimentation involves the use of 
seedlings and crop plants such as wheat, maize and bean. To some extent 
this has set a precedent so that the body of knowledge is so great for these 
model plants that further research tends to pursue research on these to 
the detriment of other species. The role of rhizosphere processes in the 
ecology of non-crop plants is almost unknown apart from a few isolated 
examples (Jones, 1998). In addition, work on trees has also been extremely 
limited, mainly focusing on temperate trees typically no more than a few 
years old due to difficulties in managing root mycorrhizal status, space 
limitations in climate control chambers and the short time horizon of much 
research (Grayston and Campbell, 1996; Grayston et at, 1997). 

Despite all the limitations on performing rhizosphere experiments, 
significant technical advances have been made in recent years which have 
allowed us to gain new insights into the biological interactions below 
ground. Although most of these techniques rely heavily on laboratory 
experimentation using either mesocosms (volume c. 50-2000 cm 3 ) or more 


D. Jones 

Soil atmosphere 







mineralization efflux 

mass flow 






sorption and 

Microbial biomass 

Soil solution 

fixation ^ 

Soil mineral particles 

. uptake 

desorption and 





mineralization \. sorption and 
N. fixation 

breakdown and 






Soil organic pool 

Fig. 15.3. Schematic representation of a mechanistic flux framework describing 
some of the possible nutrient fluxes in the rhizosphere. 

often microcosms (volume c. 5-50 cm 3 ), these should not limit their 
potential use in agroforestry research. Although this necessitates a more 
reductionist approach it also reinforces the need for large multidisciplinary 
projects and, in the case of trees, long-term research funding. Many of the 
methods outlined below were first developed for looking at biological 
interactions in bidk soil but have since been adapted for use in small soil 
volumes such as the rhizosphere. 

One of the major limitations to many rhizosphere studies performed 
in the past has been a lack of understanding of the complexity of the 
rhizosphere and an ignorance of key variables. For example it is common 
to investigate exudate release from roots without considering microbial 
consumption of the exudates (Matsumoto et al., 1979). Where experiments 
are aimed at parameterizing models, it is necessary in many cases to 
develop a mechanistic flux framework. An illustration of this is given in 
Fig. 15.3. This conceptualization provides a basic decision-making tool for 
ensuring that all factors are being considered in order for the results to be 

15.2 Obtaining a Representative Sample of Rhizosphere Soil 

Rhizosphere soil can be arbitrarily defined as the sheath of soil remaining 
attached to the roots after removal from the ground and gentle shaking. 
This offers a practical and much used approach to an almost impossible 
task. However, the amount of soil recovered can be highly dependent on 
soil and plant conditions. Soil is held to the root via a number of 

Rhizosphere Processes 


Table 15.1. Comparison between root sap and soil solution nutrient 
concentrations illustrating the need not to contaminate rhizosphere soil samples 
with plant sap. 

Solute concentration (mM) 

Root sap a 

Soil solution 15 

Fold difference 

K + 




Ca 2+ 


























Organic acids 




a Zea mays root sap (includes cytoplasm and vacuolar sap). 

b Soil solution extracted by centrifugal drainage from a free-draining Typic 


mechanisms including binding by water, fungal hyphae (mycorrhizal and 
non-mycorrhizal), adhesion with root and bacterial polysaccharides 
(mucilages) and adhesion via root hairs. Of these factors, soil moisture 
appears to largely determine the amount of soil recovered by this method, 
with little soil recovered when the soil becomes either too wet or too dry. 
There is, therefore, a critical moisture balance required for obtaining intact 
soil rhizosheaths and, where possible, soil moisture should be normalized 
across treatments. The effectiveness of the method is further reduced 
because soil adhesion is not usually uniform along the root length, making 

Mycorrhizal hypha 

Bacterial colony 

Root and bacterial 

Root epidermal ce 

Root cortical ce 


Fig. 15.4. Schematic representation of the rhizosphere zones associated with a 
vesicular-arbuscular mycorrhizal-infected root. 

296 D.Jones 

comparisons between root zones difficult. It is unusual, for example, for 
soil to adhere to the root apex. Once this stage has been achieved, the 
subsequent separation of rhizosphere soil from the root is also difficult in 
that the two are inextricably linked and recovery of one is impossible 
without some degree of contamination by the other. This is particularly 
important in soil or root samples that are being used for nutrient (organic 
and inorganic) and microbial analysis (enzymes or chemical biomass 
measures) as root sap solute concentrations are typically much greater than 
those in the soil solution (Table 15.1). Further, the most intense site of 
rhizosphere activity, the rhizoplane, which is the root surface, and the 
endorhizosphere, which is the root apoplast, are rarely recovered by this 
technique (Fig. 15.4). 

To obtain non-contaminated and carefully controlled samples of 
rhizosphere soil typically requires a scaling down of experiments to a 
mesocosm or microcosm scale, that is, to pots or rhizotrons. Although this 
does not obviate performing experiments in the field, most studies 
performed to date have been carried out within climate-controlled 
chambers, enabling dismantling in the laboratory (Marschner, 1995). 
Briefly, roots are kept separated from rhizosphere soil through a series of 
nylon or polypropylene meshes. The choice of membrane is dependent 
on root and root hair size; however, two types of mesh are usually 
employed. The first is a 0.2 |Jm pore size membrane, which is used to 
maintain a sterile root compartment and a non-sterile soil compartment. 
This is often used to determine the effects of root exudates on soil chemical 
properties or microbial biomass activity changes (Meharg and Killham, 
1991; Yeates and Darrah, 1991). A sterile root compartment is maintained 
so that exudates are not degraded by microorganisms prior to entering 
the soil. The second, more common approach involves the use of a 30-60 
|i.m mesh to separate non-sterile roots from non-sterile soil. This mesh size 
usually does not allow penetration by root hairs (diameter 50-100 |im) but 
is sufficient to allow passage of microorganisms. In addition to this, 
successive samples away from the rhizosphere can be determined with a 
resolution of up to 10 [im using a microtome approach as described by 
Yeates and Darrah (1991) and Joner et at. (1995). However, the technique 
is not without its limitations, which include: (i) interference of water and 
nutrient flow by the imposition of membranes; (ii) the exclusion of root- 
grazing mesofauna; (hi) the exclusion of the rhizoplane and 
endorhizosphere; and (iv) the difficulty of spatial isolation of individual 
root sections required for spatial analysis. Despite these limitations, 
however, this type of exclusion technique has been successfully used in 
numerous rhizosphere investigations (Gahoonia et ai, 1994; Marschner, 
1995;Alphei^a/., 1996). 

Rhizosphere Processes 297 

15.3 Methods for Rhizosphere Soil Chemistry 

Once rhizosphere and bulk (non-rhizosphere) soil samples have been 
obtained by the techniques outlined above, the chemical properties of each 
root zone can be compared by standard techniques. A typical analysis may 
include pH, major cations (ammonium, calcium, magnesium, potassium, 
sodium, aluminium), anions (nitrate, sulphate, phosphate), micronutrients 
(zinc, copper, iron) and dissolved organic carbon and nitrogen. Typically 
a variety of measures of nutrient status are performed including water or 
resin extraction, concentrated salt (KC1, NH 4 +-acetate), acid (HC1) or base 
(NaOH) extractions (Kowalenko and Yu, 1996; Trolove et al., 1996; 
Turrion et al., 1999; Zoysa et al., 1999). All of these provide information 
on differentially available fractions of soil nutrients (see Chapter 5). Soil 
solution can also be removed from larger rhizosphere samples using the 
centrifugal drainage method described by Jones and Edwards (1993). The 
details of the individual chemical analysis methods can be found in Weaver 
et al. (1994), Alef and Nannipieri (1995) and Sparks et al. (1996). 

There are now a variety of non-destructive in situ techniques for 
looking at the distribution of nutrient ions and biological activity in the 
rhizosphere. In the case of soil nutrient status, soil solution samplers have 
been designed to be able to sample gross rhizosphere solution. These are 
simply microversions of those used to sample bulk soil solution (see Section 
7.2) and can be obtained in either ceramic, polysulphone or other organic 
materials (Gottlein et al., 1996; Farley and Fitter, 1999). Although these 
do allow an estimate of rhizosphere solution chemistry to be made, the 
sphere of sampling is often unknown, they only operate in moist soils (< 0. 1 
MPa) and they recover only small sample volumes. The problems of 
sample volume can be overcome through the use of inductively coupled 
plasma mass spectrometry (ICP-MS), but predicting the sampling volume 
is extremely difficult. Alternative more spatially specific methods involve 
the use of fine glass capillaries, which allow the extraction of extremely 
small sample volumes (10 pi), which can subsequently be analysed by 
capillary electrophoresis with a reasonable degree of resolution (> 10 u.M) 
(Bazzanella et al., 1998). This technique is by far the most powerful; 
however, it requires micromanipulators, capillary pulling equipment and 
good dissecting microscopes and is labour intensive. 

The overlaying of agar or filter papers impregnated with colorimetric 
indicators on to exposed roots has also been widely employed for 
determining rhizosphere nutrient dynamics. This technique is ideal for 
semiquantitative studies in which spatial dynamics are particularly 
important, for example when distinguishing effects of the root tip, root 
hair zone and root base. If the colorimetric indicators are non-rhizotoxic, 
then multiple exposures can be performed over time; however, in most 
cases only single time point measurements have been recorded. This 

298 D.Jones 

technique has been used widely for determining the spatial localization of 
pH changes in the rhizosphere and especially in response to changes in 
nitrogen and phosphorus nutrition. Other applications involve the study 
of rhizosphere phosphatase activity, root ferric reductase activity, Mn 4+ 
reduction, solubilization of inorganic phosphate and the release of organic 
acids (Dinkelaker et al. , 1993a,b). 

Oxygen microelectrodes have also been developed which again are 
capable of non-destructively determining rhizosphere O, concentration in 
a mesocosm-type system (Revsbech et al, 1999). The advantages and 
disadvantages are discussed in Armstrong (1994). In addition, an oxygen- 
sensing reporter strain of Pseudomonas fluoresceins has also been used for 
monitoring the distribution of low-oxygen habitats in soil (Hojberg et al. , 
1999), as have CH 4 -sensitive biosensors (Damgaard et al, 1998). 

15.4 Methods for Rhizosphere Biological Activity 

In recent years, great advances have been made in soil microbial ecology, 
allowing not just the biochemical characteristics of soils to be determined, 
such as enzyme activity, but also the size, activity and community structure 
of the rhizosphere microbial population. When investigating biological 
activity in the rhizosphere it is advisable to adopt a multifaceted approach. 
Typically this will involve a basic measure of soil microbial biomass (Beck 
et al., 1997) and microbial activity, usually basal or substrate-induced 
respiration (Jensen and Sorensen, 1994), alongside standard measures of 
key soil enzymes involved in carbon, nitrogen and phosphorus cycling 
such as protease, deaminase, cellulase and phosphatase activity (Alef and 
Nannipieri, 1995; Naseby et al., 1998). Although these techniques are 
straightforward and can be performed easily in most laboratories, 
measures of soil microbial community structure often require specialist 
equipment. At present a number of approaches can be taken, including 
the use of restriction fragment length polymorphism analysis (RFLP) 
(Chelius and Lepo, 1999), randomly amplified polymorphic DNA analysis 
(RAPD) (Redecker et al, 1999), fatty acid methyl ester analysis (FAMES) 
(Olsson and Persson, 1999; Siciliano and Germida, 1999), 16S rRNA 
amplification and temperature gradient gel electrophoresis polymerase 
chain reaction (TGGE-PCR) (Schwieger and Tebbe, 1998; Heuer et al., 
1999; Smit et al, 1999), Biolog™ (Germida and Siciliano, 1998), 
enterobacterial repetitive intergenic consensus sequence PCR (ERIC-PCR) 
(DiGiovanni et al, 1999) and phospholipid fatty acid profiling (PLFA) 
(Griffiths et al, 1999). The most convenient of these techniques is Biolog™ ; 
however, it provides information only on culturable microorganisms and 
has therefore been criticized (Howard, 1997; Lawley and Bell, 1998). 

Rhizosphere Processes 299 

Although rhizosphere biological activity can be determined in 
destructively sampled rhizosphere soil, this typically provides little spatial 
or temporal resolution. This may be especially important considering that 
typically only 10% of the rhizoplane is colonized and that where 
colonization does occur it can be highly dependent on root architecture, 
such as cell junctions and the endorhizosphere. However, new techniques 
are now available which allow microbial activity and biomass of specific 
species to be determined in situ and in vivo. Further, the spatial and 
temporal dynamics can also be determined with high accuracy. The success 
of these techniques, however, relies on light emission from samples and 
therefore experimental design is critical to their success. For this reason, 
experiments are typically conducted in small microcosms and can only be 
performed in the laboratory. 

One of the main approaches has been to genetically modify target 
organisms with a bioluminescent luciferase (luxAB) gene, followed by the 
introduction of the tagged organism into microcosms. The amount of light 
produced by organisms established in the rhizosphere can subsequently 
be detected at a gross scale using commercial luminometers or with 
reasonable spatial resolution (0.05-10 cm) using a CCD-cooled camera 
(Deweger et al. , 1991; Rattray et al. , 1995; Prosser«/!a/., 1996). The amount 
of light produced by the tagged organisms can be correlated with microbial 
activity, allowing target organism activity as a function of time and space 
to be determined (Eberl^ al, 1997; Hagen^ al, 1997; Wood etal., 1997). 
In addition, the luxAB gene insert can be placed into the target organism 
behind specific gene promoters, allowing the expression of individual 
microbial metabolic pathways to be assessed, such as those involved in 
nitrogen and phosphorus starvation (Kragelund etal., 1997; Jensen et al., 
1998; Milcamps et al., 1998). 

Another similar approach has been the introduction of micro- 
organisms tagged with a green fluorescent protein (GFP) into rhizosphere 
soil. This allows detection of introduced microorganisms using standard 
epifluorescence and confocal microscopes. The constitutively expressed 
GFP can be introduced as a quasi-stable tag, allowing total population 
numbers to be evaluated. In contrast, the introduction of an unstable GFP 
tag allows the expression of specific genes and metabolic pathways to be 
determined, thereby providing measures of microbial activity (Gage et al., 
1996; Normander et al., 1999; Ravnskov et al., 1999; Tombolini et al., 
1999). In a similar fashion to that of the luxAB genes, the GFP cassette can 
also be placed behind specific promotors, allowing the study of specific 
metabolic pathways such as those involved in nitrogen fixation (Egener 
etal., 1998). 

300 D.Jones 

15.5 Quantification of Root Carbon Loss into the 

The driving force for many rhizosphere processes is carbon loss from the 
root. Quantification of this carbon loss can be achieved in a variety of ways. 
The first involves whole-plant radiolabelling by continuous or pulse 
feeding the leaves with 14 C0 2 either in the laboratory or in the field, as 
described by Meharg and Killham (1988). Although this approach allows 
total plant carbon budgets to be calculated it does not provide information 
on the type of carbon compounds being released and has been criticized 
(Meharg, 1994). To determine the spectrum of exudate loss, a number of 
approaches can be taken, including extraction of rhizosphere soil solution 
by centrifugal drainage (Strom, 1997), the collection of charged exudate 
compounds by ion-exchange resins (Kamh et ai, 1999) and the growth of 
plants under sterile conditions followed by collection in hydroponic or 
sand culture (Jones and Darrah, 1993; Hodge et ai, 1998). The problem 
associated with determining carbon loss in sterile systems is discussed in 
Jones and Darrah (1993). The analysis of exudate components is usually 
carried out by standard gas chromatography, capillary electrophoresis, 
high-pressure liquid chromatography and enzymatic procedures, details 
of which can be found in any standard biochemical methods text. In 
addition, sites of carbon loss can also be determined semiquantitatively 
with bacterial biosensors (Jaeger et al. , 1999). 

15.6 Rhizosphere Mathematical Modelling 

Mechanistically based computer simulation models are available to predict 
the spatial and temporal dynamics of complex rhizosphere processes (see 
Fig. 15.1). They incorporate parameters such as diffusion and mass flow 
of nutrients and water, nutrient sorption/desorption and fixation reactions, 
soil solution chemical equilibria and characteristics of root nutrient influx, 
root radius, root hairs and shoot demand. Most of the models developed 
have been used for predicting nutrient uptake at the root, whole plant and 
field level (Sadana and Claassen, 1999). The construction of these models 
and examples of their output are extensively described in Tinker and Nye 
(1999), Cushman (1984) and Barber (1995). Other types of models have 
been developed from this basic design into which redox reactions have 
been incorporated (for rice-paddy soils) or for work in model rhizosphere 
systems (Jones and Darrah, 1993; Kirk and Solivas, 1994; Calba et al., 
1999). The latest generation of models has been used to describe bacterial 
dynamics in the rhizosphere in response to root carbon additions (Darrah, 
1991b; Scott et al, 1995), whereas others have fused nutrient and bacterial 

Rhizosphere Processes 301 

models to describe nutrient cycling involving biodegradable nutrient- 
mobilizing organic ligands (Geelhoed et a!,., 1999; Kirk, 1999; Kirk et al, 
1999). For a recent review see Toal et al. (2000). 

Chapter 1 6 
Soil Macrofauna 

P. Lavelle, 1 B. Senapati 2 and E. Barros 3 

1 Laboratoire d'Ecologie des Sols Tropicaux, 32 rue Henri Varagnat, 
93143 Bondy Cedex, France; 2 Ecology Section, School of Life 
Sciences, Sambalpur University, Jyoti Vihar, 768019, Orissa State, 
India; 3 National Institute for Research in the Amazon (INPA), CP 
478, 69011-970 Manaus, AM, Brazil 

16.1 Synopsis 

There is growing awareness that the sustainability of agricultural systems 
will largely depend on the adequate management of biological resources 
(Woomer and Swift, 1994). In the suite of factors that determine soil 
fertility, soil organisms are proximate regulators of nutrient availability 
and soil structure (Lavelle, 1997). Their importance as a resource for 
agriculture has been underestimated because often only constraints 
imposed by climate and chemical parameters of soil fertility have been 
considered as aspects that could be influenced by management. Since the 
1980s and the new soil fertility paradigms proposed by the Tropical Soil 
Biology and Fertility Programme (TSBF), emphasis has been placed on 
the cultivation of plant species that ensure nitrogen fixation and provide 
organic matter and nutrients for the system while generating sufficient 
income to sustain farmers' livelihoods (Sanchez, 1994; Woomer and Swift, 
1994). In addition to plant leaf litter and root systems, whose important 
roles in carbon and nutrient cycling, maintenance of soil structure and 
biological activity have been discussed in previous chapters of this book, 
soil invertebrates constitute another biological component that regulates 
basic soil processes such as soil aggregation, porosity and organic matter 
dynamics. Their management is now considered important, especially 
because most current land-use practices either destroy soil faunal 
communities or severely deplete their diversity, often with negative 
consequences for soil fertility (Chauvel et ai, 1999; Lavelle et at., 1999). 
In this chapter, the composition and functional structure of soil 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 303 

F.L. Sinclair) 

304 P. Lavelle etal. 

invertebrate communities are discussed, their effects on soil fertility are 
explained in broad terms, the effects of agroforestry practices on their 
communities are analysed, and techniques to assess their composition, 
abundance and activity are described. 

Soil organisms and their functional domains 

Soil organisms have evolved in an environment that imposes three major 
requirements: (i) to move in a compact environment with a loosely 
connected porosity; (ii) to feed on low-quality resources; and (hi) to adapt 
to the occasional drying or flooding of the porous space (Lavelle, 1997). 
A continuum of adaptive strategies based on size is observed in soils, from 
microorganisms to macroinvertebrates (Swift et al., 1979). Soil 
invertebrates have been divided into micro-, meso- and macrofauna 
depending on the average size of the individuals. Microfauna comprises 
invertebrates <0.2 mm that live in the water-filled porosity of soil; 
mesofauna includes invertebrates 0.2-10 mm in length that live in the air- 
hlled soil porosity and in the litter; and macrofauna comprises the largest 
invertebrates (> 1 cm). In more general terms, a soil macrofauna taxon 
may be defined as an invertebrate group found within terrestrial soil 
samples which has more than 90% of its specimens in such samples visible 
to the naked eye (P. Eggleton et al., unpublished). According to this 
definition, large individual Collembola or Acari should not be considered 
as macrofauna. 

Microorganisms have developed the ability to digest every substrate 
present in soils. Unable to move, they depend on other organisms to 
transport them to where they can come into contact with the substrates on 
which they feed. Soil invertebrates have limited abilities to digest the 
complex organic substrates of the soil and litter, but many have developed 
interactions with microflora that allow them to exploit soil resources. With 
growing size, the relationship between microflora and fauna gradually 
shifts from predation to mutualisms of increasing efficiency. Excrement of 
invertebrates is of utmost importance in the evolution of organic matter 
(Martin and Marinissen, 1993), in the formation and maintenance of soil 
structure and, over long periods of time, in specific pedological processes 
referred to as zoological ripening of soils (Bal, 1982). Three major guilds 
of soil invertebrates, described below, may be distinguished on the basis 
of the relationships that they have with soil microorganisms and the kind 
of excrement and other biogenic structures that they produce. 

• Micropredators mainly comprise microfauna that feed on bacteria, 
fungi and other micropredators. Microfauna do not seem to produce 
recognizable solid excrement and so their effect on soil organic matter 

Soil Macrofauna 305 

dynamics is not prolonged in structures that are stable for some time 
after deposition. They do, however, have a significant impact on the 
population dynamics of microorganisms and the release of nutrients 
immobilized in microbial biomass (Trofymow and Coleman, 1982; 
Clarholm, 1985). This process is especially developed in the 
rhizosphere (see Chapter 15). Predatory Acari or Collembola and even 
larger invertebrates (earthworms) may extend this food web over 
several trophic levels (Hunt, 1987; Moore et al, 1993). 

• Litter transformers mainly comprise mesofauna and large arthropods 
that normally ingest purely organic material and develop an external 
(exhabitational sensu; Lewis, 1985) mutualism with microflora based 
on an external rumen type of digestion (Swift et al, 1979). Litter 
arthropods may digest part of the microbial biomass or develop 
mutualistic interactions in their faecal pellets. In these structures, 
organic resources which have been fragmented and moistened during 
the gut transit are actively digested by microflora. After some days of 
incubation, arthropods often re-ingest their pellets and absorb the 
assimilable organic compounds that have been released by microbial 
activity, and occasionally part of the microbial biomass itself (see, for 
example, Hassal and Rushton, 1982; Szlavecz and Pobozsny, 1995). 

• Finally, the ecosystem engineers comprise macrofauna, mainly 
earthworms, termites and ants, that are large enough to develop 
mutualistic relationships with microflora inside their proper gut. 
These organisms usually ingest a mixture of organic and mineral 
elements. Large faecal pellets (in the range of 0. 1-2 cm and more) may 
be the component elements of macro-aggregate structures and 
participate prominently in the formation of stable structures through 
the regulation of porosity, aggregation, bulk density and surface 
features (Bal, 1982; Blanchart^a/., 1999; Decaensrfa/., 1999). These 
organisms also build large structures, such as mounds and networks 
of galleries and chambers, which have significant impacts on the 
evolution of soils over medium time scales. Ants may be considered 
part of this group, although the vast majority of them only use soil as 
a habitat and have a limited impact on soil organic matter dynamics. 

Whenever conditions are suitable for their activities, macroinvert- 
ebrates, and especially earthworms and termites, become major regulators 
of microbial activities within their sphere of influence, referred to as the 
termitosphere of termites and the drilosphere of earthworms (Lavelle, 
1997). Here, they also determine the abundance and activities of smaller 
groups of soil fauna (Dashetal., 1980; Yeates, 1981; Boyer, 1998; Loranger 
et al, 1998). These functional domains also include the rhizosphere, in 
which roots are the major determinants of microbial and faunal activity 
(see Chapter 15). 

306 P. Lavelle etal. 

Communities of organisms in soil are thus largely interactive, with a 
clear hierarchical organization dominated by ecosystem engineers. This 
justifies putting special emphasis on the management of macroin vertebrate 
communities for manipulating the whole biological activity of soils that 
they actually regulate inside their functional domains. The sampling of 
macroin vertebrate communities, which comprise most ecosystem 
engineers and litter transformers, is therefore proposed as a suitable proxy 
for overall soil biological activity (see Section 16.2). 

Attempts have been made to further classify macroinvertebrates into 
functional groups with a real functional meaning (Bouche, 1977; Grasse, 

Box 16.1. Physical and chemical properties of biostructures 
deposited by soil invertebrate engineers at the soil surface 
in the eastern Llanos of Colombia. 

In a natural savanna of eastern Colombia, structures deposited at the soil surface 
by 1 7 species of soil invertebrate engineers were analysed for their chemical and 
physical characteristics and data were treated with principal component analysis 
(PCA) (Fig. 1 6.1 ). The analysis separated structures into three groups corresponding 
to large taxonomic units (earthworms, ants and termites). Earthworm casts were 
characterized by high values of bulk density, pH and contents of organic matter 
and cations. Termite structures differed from earthworm casts in having lower pH, 
lower bulk density and larger diameter, which reflects a greater structural stability. 
Ant structures had the highest aluminium saturation and the lowest amounts of 
soil organic matter and cations. This study confirms that physicochemical 
properties of biostructures are related to the faunal taxonomic units that produce 
them, suggesting that these represent, to a certain extent, functional groups. An 
exception was the group of Termitidae sp. 4, whose sheath ings were similar to ant 
surface nest material (Decaens etal., 2001). 

Fig. 16.1. (Opposite) Biostructures deposited on the soil surface of the eastern 
Llanos of Colombia by 1 7 species of soil invertebrate engineers as ordinated 
with principal component analysis (PCA) according to their physical and 
chemical properties: (a) Eigen values for extracted factors (the first two axes 
represent 59.6 and 22.6% of total inertia, respectively); (b) correlation circles 
associated with the first two axes; (c) ordination of biostructures in the plane 
delimited by Factors 1 and 2 of the PCA (small circles are means of points 
representing a category of structures produced). Abbreviations: %Corg, per cent 
organic carbon; PBrayll, available phosphorus content (Brayll); Ntotal, total 
nitrogen; Ptotal, total phosphorus; Al, exchangeable aluminium; Ca, 
exchangeable calcium; Mg, exchangeable magnesium; K, exchangeable 
potassium; %Al, per cent aluminium saturation; BD, bulk density; MWD, mean 
weighted diameter of aggregates; V, earthworm casts; T, termites (nests of all 
species except for surface sheathings of Termitidae sp. 4); A, ant surface nests 
(reproduced with permission from Decaens etal., 2001). 

Soil Macrofauna 


K Ptotal 
" Ntotal 

Camponotus sp. 

Termitidae sp. 1 

Nasutitermitinae sp. 

Termitidae sp. 3 

> ^ I 

\ Acromyrmex landolti 
Trachymyrmex sp. 

Axis 1 


Martiodrilus carimaguensis 

V J Andiodrilus yoparensis 

1984). The effects of disturbances or changes in the environment on 
invertebrates indicate that large taxonomic units (such as termites and 
earthworms) are often fairly homogeneous groups in terms of response 
(sensu; Lavorel et ai, 1997) and represent a rough approximation to real 
functional groups, that is, groups of species that perform similar functions 
within ecosystems (Steneck, 2001). For example, in natural savannas of 

Table 16.1. Functional categories of earthworms. 1 









Litter anc 




A horizo 


A and B 

Litter feeder Fully pigmented 

Litter and soil Feeds on litter and soil Anterodorsally 

Rich soil feeder Nil 

Bulk A, soil feeder Nil 

Deep soil feeder 



Nil or very limited 



Long-lasting vertical 

Medium to large 


Horizontal, partly 
refilled with casts 



Extensive horizontal 
galleries refilled with 



Horizontal galleries 
refilled with casts 

Large to very large 

a After Bouche (1977) and Lavelle (1983). 

Soil Macrofauna 309 

Colombia, the physical and chemical properties of biostructures produced 
by 17 different mac roinverteb rate species were clearly related to large 
taxonomic units, with only one exception (Decaens et at. , 2001; see Box 16.1, 
p. 306). 

Within large taxa, species can be further subdivided into functional 
groups of species with distinct roles in the soil. Earthworms have been 
classified into five different categories depending on their feeding habits 
and location in the soil (Table 16.1; Lavelle, 1983). These categories have 
different distinct effects on soil function. Epigeics are useful agents in the 
production of compost but poor bioturbators (i.e. organisms that mix litter 
and soil). Anecics create persistent vertical below-ground networks of 
galleries that open at the soil surface; they also produce large numbers of 
casts, which they deposit in soil macropores or at the soil surface. Endogeics 
are the most active bioturbators; they deposit most of their casts in the soil 
and dig horizontal below-ground galleries that they refill with their casts. 
They have been further divided into poly-, meso- and oligohumics with 
decreasing average concentrations of organic matter in the ingested soil. 

Termites may be broadly classified on the basis of their feeding habits, 
general organization of their nests, and their location in the environment. 
Nests may be located in five main sites: (i) within the wood of living and 
dead trees and in fallen timber; (ii) subterranean nests; (iii) epigeal nests 
on the soil surface; (iv) arboreal nests hanging from branches or stuck to 
tree trunks; and (v) within the nests of other termite species (Noirot, 1970). 
Termites may also be separated into four groups on the basis of their 
feeding habits: (i) wood feeders; (ii) fungus cultivators; (iii) grass harvesters 
and surface litter feeders; and (iv) humivorous termites (Josens, 1983). A 
small number of species also regularly attack living plants. Wood feeders 
may be further separated into several groups, such as dry-wood, wet-wood 
or arboreal termites (Abe, 1987). 

No specific classifications have been proposed so far for other soil and 
litter macroin vertebrates. A classification of ants based on their effects on 
bioturbation and transfers of organic matter in soil is clearly needed. Litter 
transformers (mainly Diplopoda, Isopoda and detritivore Coleoptera) 
seem to be a rather homogeneous functional group, although they may 
differ significantly in their response to environmental factors. Predators 
are mostly found in Myriapoda (Chilopoda), Insecta (Coleoptera and ants) 
and Arachnida. 

Effects of macroinvertebrates on soil processes 

Fauna effects on soil physical properties 

Soil invertebrates influence soil processes directly by their digestion 
processes and the formation of biogenic structures. The latter effect is by 

310 P. Lavelle etal. 

far the most important since it determines basic soil physical properties 
and further steps of organic matter dynamics. Soil aggregation in most 
humid tropical environments is determined by the activity and functional 
diversity of the main ecosystem engineers. In moist savannas of Cote 
d'lvoire, earthworms may produce up to 1200 t of casts ha 1 year 1 , of 
which only 30 t (2.5%) are deposited on the soil surface (Lavelle, 1975). 
The earthworm community comprises compacting species that ingest small 
aggregates and produce large compact structures, the accumulation of 
which tends to increase the bulk density of the soil (Blanchart et al, 1999), 
and decompacting species, which have opposite effects through the 
fragmentation of large aggregates into smaller ones and/or the creation of 
pores. The combined effects of these two broad functional categories 
maintain soil physical and hydraulic properties in a highly dynamic state. 

Ants and termites also have strong impacts on soil physical parameters 
(Lobry de Bruyn and Conacher, 1990). Some Macrotermitinae may open 
access holes at the surface, which they use for foraging at night. The overall 
surface of these openings has been estimated at 2-4 m 2 ha _1 in a dry savanna 
of Kenya (Lepage, 1981). Gallery lengths of up to 7.5 km ha _1 have been 
estimated for soils associated with the mounds of Macrotermes michaelseni 
(Darlington, 1982; MacKay et al, 1985; MacKay and Whitford, 1988). In 
the soils of Darlington's (1982) study area there was also the equivalent of 
90,000 storage chambers per hectare. It can be calculated that the voids 
formed by termite galleries and related structures occupied approximately 
0.4% of the soil volume to 20 cm depth at this site (Lavelle and Spain, 2001). 

The effect of soil fauna on soil structure can be surprisingly rapid. 
Barros et al. (2001) showed that the structure of an Amazonian Oxisol 
could be drastically modified by changes in faunal activity in only 1 year. 
Soil monoliths taken from a pasture soil that had been severely compacted 
by an invasive earthworm were transported to a nearby forest. After 1 year 
of exposure to the diverse fauna of the forest, soil of the exposed monoliths 
had recovered the overall porosity and the percentage of macroaggregates 
had returned to the situation found in forest soil. 

While this example demonstrates the importance of decompacting 
fauna species for the maintenance of soil hydraulic properties, compacting 
species may also play an important role through the stabilizing effect of 
the large and dense aggregates that they produce on soil structure which 
protects against mechanical stress and maintains organic matter in the 
interior of these dense aggregates (see below and Martin, 1991). 
Furthermore, an earthworm species that compacts the superficial soil 
horizons may increase soil porosity in the subsoil (Barros et al., 2001). The 
dynamic equilibrium between these antagonistic activities has a strong 
influence on soil physical, chemical and biological properties, and 
illustrates the importance of the functional diversity of soil organisms for 
soil functioning and fertility. 

Soil Macrofauna 311 

Biogenic structures produced by ecosystem engineers are often rather 
resistant, especially after having experienced a few wetting and drying 
cycles. This is the case for earthworm casts that are highly unstable when 
fresh and soon acquire high stability after drying (Shipitalo and Protz, 
1988; Blanchart et al, 1993; Hindell et ai, 1997). Dry surface casts of 
Martiodrilus carimaguensis, a large anecic earthworm of the Colombian 
Llanos Orientales, can resist heavy rainfall and persist for several months 
before rainfall and runoff break them down into stable aggregates that 
form a mulch on the soil surface (Decaens, 2000). Unless they are 
destroyed by another invertebrate, biogenic structures created inside the 
soil may persist for long periods of time after the organisms that produced 
them have died. Process regulation by ecosystem engineers may therefore 
persist long after they have disappeared, and degradation of soil properties 
due to their elimination by cropping practices, or any other disturbance, 
is often ignored because the link between the organism and the soil 
properties is not evident. 

Soil texture in the upper soil horizons may also be modified by 
macroin vertebrate activities as a result of the regular deposition of finely 
textured biogenic structures at the surface. This process has been 
hypothesized to generate stone-lines found at some depth in many tropical 
soils (Williams, 1968; Wielemaker, 1984; Johnson, 1990). In other 
circumstances, fine particles deposited by fauna at the soil surface may be 
washed away by runoff and leave a coarse-textured surface horizon (see 
Chapter 17). 

Fauna effects on soil organic matter dynamics 

The effects of soil invertebrates on soil organic matter dynamics are 
disproportionate to their size and numbers. Direct mineralization of soil 
organic matter by earthworms was estimated at 1.2 t ha -' year 1 in a natural 
savanna of the Cote d' I voire (Lavelle, 1978). This corresponds to 
approximately 10% of all carbon mineralized in this soil, and the overall 
contribution of all soil fauna to carbon mineralization at this site probably 
does not exceed 20% (Lamotte, 1989). Indirect effects on organic matter 
dynamics are much larger, especially as they extend over several time 
scales. At a short time scale of days to weeks, the organic matter that has 
been ingested by invertebrates and egested in their faecal pellets 
undergoes further mineralization caused by a strong enhancement of 
microbial activity, labelled the 'Sleeping Beauty effect' by Lavelle (1997). 
At a longer time scale of months to years, organic matter occluded in 
compact biostructures like earthworm casts or walls of termite mounds 
may be physically protected from decomposition. Therefore, invertebrates 
that stimulate mineralization of organic matter at small scales of time and 
space tend to protect organic matter from further decomposition at much 

312 P. Lavelle etal. 

larger scales. The coexistence of these two processes interacting at distinct 
scales may lead to misinterpretation of some experimental results. 
Elimination of invertebrate populations only suppresses short-term effects, 
whereas long-term protection of organic matter in biogenic structures will 
last as long as these structures are maintained. However, Blanchart et al. 
(2003) have shown that this regulation of soil organic matter dynamics via 
the effects on soil physical structure was only significant in soils with low 
clay contents and soils with inactive clay minerals of the 1 : 1 type, but not 
in soils with smectites and other 2:1 clay minerals. 

Effects of agroforestry practices on soil macrofauna 

Hierarchical determinants of soil function in agroforestry systems 

Soil function is controlled by a suite of hierarchically organized 
determinants, the relative importance of which depends on the scale at 
which they operate (Lavelle and Spain, 2001). Agroforestry practices will 
affect soil function as they act on these determinants. At the top of the 
hierarchy are climatic factors such as moisture and temperature regimes. 
At the next level down are edaphic parameters with special importance 
accorded to the nature and abundance of clay minerals and the overall 
nutrient richness of soil, which is partly determined by the nature of the 
original bedrock. At the next level down comes the quality of organic 
matter produced and other parameters linked to the plant communities 
that are present. Macroin vertebrate communities are dependent on these 
determinants and constraints imposed by biogeographical factors, such as 
the local history and spatial distribution of land-use systems, which 
impinge on the maintenance of diversity and colonization of newly created 
cropping systems from adjacent areas. These communities, in turn, are 
themselves a proximate determining factor for microbial activities in the 

The few available data sets reflect this diversity of determinants of soil 
invertebrate communities (Lavelle and Pashanasi, 1989; Lavelle et at. , 1994; 
Loranger, 1999). Agroforestry systems appear to have rather specific fauna 
communities compared with conventional cropping systems. Trees in 
agroforestry systems affect living conditions of invertebrates by their shade, 
their deep and perennial root systems, and the quantity and quality of 
their litter (Tian et al., 1995a; Vohland and Schroth, 1999), and by 
preventing soil tillage in at least part of the system. 

Tillage has a significant negative impact on soil macroin vertebrates. 
There is a direct effect through disruption of the soil structure and the 
destruction of either the invertebrates themselves or the structures that they 
inhabit. In the longer term, a combination of microclimatic and trophic 

Soil Macrofauna 313 

effects prevents populations from building up rapidly since soil is left bare 
for some time after each tillage (House, 1985; Haines and Uren, 1990). 

Soil microclimate is affected by agroforestry practices in two different 
ways. Soil under trees is covered with leaf litter and shaded, so that the 
upper centimetres of soil are protected from overheating and drying (Lai, 
1986). On the other hand, the perennial nature of trees, their often high 
evapotranspiration and the rather deep distribution of roots may cause 
more intensive and deeper drying of the soil during the dry season than 
in areas with annual crops or other herbaceous vegetation, and this may 
affect soil fauna. 

The quality and abundance of organic inputs is a key feature in the 
maintenance of diverse soil macroin vertebrate communities (Lavelle et al. , 
2001). The provision of habitats and food for litter invertebrates favours 
the maintenance of specific communities of epigeic invertebrates (mainly 
arthropods such as Myriapoda, Coleoptera, ants or surface termites, and 
epigeic earthworms) and anecic invertebrates which dwell in the soil but 
feed, at least partly, on surface leaf litter, such as termites and earthworms. 
Barros et al. (1999) showed an effect of litter cover on soil fauna in a 
comparison of two different agroforestry systems and three 10-year fallows 
in central Amazonia. In a system with still young trees, litter cover was low 
and the macroinvertebrate community had a lower overall abundance and 
diversity and tended to live deeper in the soil than in the other sites with 

Fig. 16.2. Relationships between litter dry matter, faunal density (a) and faunal 
biomass (b) for different sampling positions in a multistrata agroforestry system 
with Brazil nut trees (Bertholletia excelsa), cupuacu (Theobroma grandiflorum), 
annatto (Bixa orellana), peach palm (Bactris gasipaes), a Pueraria phaseoloides 
cover crop and spontaneous grass as well as a peach palm monoculture in central 
Amazonia (means and s.e.; reproduced with permission from Vohland and 
Schroth, 1999). 

314 P. Lavelle etal. 

older trees and more abundant litter. In the same region, Vohland and 
Schroth (1999) observed a close relationship between the amount of litter 
and the abundance of litter-dwelling macroin vertebrates. They estimated 
the amount of litter necessary for maintenance of an active fauna in the 
litter system at 3-6 t ha -' (Fig. 16.2). 

The amount of soil organic matter which is assimilated by an 
earthworm community in a natural savanna of Cote d' I voire has been 
estimated at 1.2 t ha _1 year 1 (Lavelle, 1978). Based on this number, one 
could assume that if agroforestry practices succeeded in supplying 2 t ha _1 
year 1 of assimilable organic matter this should be enough to sustain 
macrofaunal activity at a suitable level. We do not know, however, how 
much litter of what quality must be applied to the soil surface to sustain a 
litter biomass of 3-6 t ha -1 as a habitat for litter-dwelling invertebrates while 
providing 2 t ha -1 year- 1 of organic matter for digestive assimilation. This 
question needs to be resolved for various agroforestry practices on 
different sites, taking the quantity and quality of available organic materials 
into consideration. 

Assimilability of plant residues is greatly determined by their chemical 
composition, especially lignin, polyphenol and nitrogen contents (see 
Chapter 6), and this influences the composition of the faunal community. 
In general, the ratio of termite to earthworm abundance is dependent on 
the quality of the organic residues; termites tend to predominate when 
low-quality organic substrates are available whereas earthworms thrive 
better with high-quality residues. In rubber plantations of southern Cote 
d'lvoire, for example, xylophagous and other termites dominate in the 
early phases when trunks and branches of the original forest are left on 
the soil surface. At later stages, 10-20 years after plantation establishment, 
most woody litter has been eaten up and transformed by termites, and 
endogeic earthworms, which presumably feed on termite excreta, become 
dominant (Gilot et al., 1995). Tian et al. (1993) showed similar effects with 
different mulch covers in agroforestry systems in Nigeria. 

In silvopastoral systems, the use of improved pastures with selected 
African grasses and legumes, as well as grazing, generally increases 
earthworm abundance and the ratio of earthworm to termite population 
densities (Decaens et al., 1994; B. Senapati, unpublished data). Cover crops 
and grass fallows may have similar impacts depending on the quality of 
organic matter produced. In a survey of soil macroinvertebrate 
communities in the Brazilian Amazon, E. Barros et al. (unpublished) found 
that mean values of the termite:earthworm density index, calculated from 
population densities, ranged from 0.2 in pastures to 7.9 and 8.8 in 
secondary forests and agroforestry systems and 20.4 and 2 1.4, respectively, 
in fallows and crops. The similarity between fallows and crops in this case 
indicates that the fallows were still young and so there was relatively little 
production of leaf and woody litter by the trees. Vohland and Schroth 

Soil Macrofauna 315 

(1999) found that in Amazonian agroecosystems the fleshy harvest residues 
of peach palm (Bactris gasipaes), grown for heart-of-palm production, 
harboured diplopods and other fauna groups sensitive to desiccation, 
which, in the primary forest of the region, are typically found in rotting 
stems of fallen trees. 

Tree species effects within agroforestry systems 

The distribution and nature of plants in an agroforestry system will greatly 
influence macro in vertebrate communities. The particular trophic and 
microclimatic conditions in the proximity of trees may significantly shape 
the invertebrate community and lead to single-tree effects on their 
distribution (Blair etal., 1990; Boettcher and Kalisz, 1991). In a rainforest 
of French Guiana, endogeic earthworms were concentrated below the litter 
of an unidentified tree species of the Qualea genus but were totally absent 
from litter of another tree, Dicorynia guianensis, only 20 m away (Wardle 
and Lavelle, 1997). Distribution patterns of litter fauna related to the 
presence of individual trees have also been observed in a multistrata system 
with four tree crop species, a leguminous cover crop and spontaneous 
grasses in central Amazonia (Vohland and Schroth, 1999). Some fauna 
groups showed specific preferences for certain litter types. For example, 
earthworms were most abundant in the litter of the cover crop, Pueraria 
phaseoloides, peach palm {Bactris gasipaes) and annatto (Bixa orellana), but 
were absent from the hard and rather dry litter of cupuagu (Theobroma 
grandiflorum) and Brazil nut (Bertholletia excelsa). Similar preferences were 
shown by millipedes and beetles (Fig. 16.2). In another multistrata system 
established after pasture in the same region, Barros et at. (1999) found a 
dominance of mesohumic endogeic earthworm species in the soil under 
cupuacu and Brazil nut trees and of polyhumic earthworm species under 
mahogany (Swietenia macrophylla) and passion fruit (Passiflora edulis) within 
the same system. Presumably, this difference also reflected the low litter 
quality of cupuagu and Brazil nut and, for young cupuagu trees, the small 
quantity of litter produced. 

At a landscape scale the distance from an agroforestry plot to areas of 
natural vegetation (where it still exists) and the influence of neighbouring 
types of land use are expected to affect the composition of fauna 
communities. In central Amazonia, the colonization of agroforestry plots 
by a large anecic glossoscolecid earthworm seemed to depend on contact 
with natural forest (E. Barros, unpublished). However, adequate data sets 
to properly test this hypothesis are lacking. 

The role of biodiversity 

Agroforestry systems are generally considered to have a positive effect on 
the conservation of biodiversity compared with simpler agricultural 

316 P. Lavelle etal. 

systems (Fragoso and Rojas, 1994; Griffith, 2000). This is especially the 
case when their structure is close to that of the original ecosystem (Decaens, 
2000). Conservation of biodiversity is important both from a general ethical 
point of view and because it may be a precondition for the sustainable use 
of soils. However, this latter point is still poorly understood and limited to 
a few hypotheses (Hooper et a!,., 2000; Lavelle, 2000). It is assumed that 
below-ground faunal diversity is influenced by the diversity of plants 
growing at a site and vice versa. However, the relationship is complex since 
a large number of interactive processes are involved. Little is known about 
the role of diversity in soils under field conditions beyond experimental 
evidence that a minimum diversity directly influences rates of basic 
ecosystem processes such as mineralization of carbon and nutrients (Mikola 
and Setala, 1998). The relationship between species richness and 
functional diversity seems to be rather loose since some groups, like spiders 
or Coleoptera, may be locally represented by more than 100 species with 
high apparent functional redundancy whereas termites would only have 
20-40 species and earthworms approximately ten species. The latter two 
groups may actually occupy much wider niche spaces than non-social 
arthropods because of their social organization and/or ecological plasticity 
(Lavelle and Spain, 2001). In some cases, significant degradation of soils 
has been related to an imbalance between functional groups, such as 
invasion of an Amazonian pasture by a peregrine earthworm species, 
which accumulated large amounts of compact structures close to the soil 
surface and, in combination with depletion of decompacting species (ants 
and termites), led to the formation of a continuous surface crust, which 
impeded plant growth (see above and Chauvel et ai, 1999). The density 
of the compacting earthworm species was much lower in a small woody 
area (E. Barros and T. Decaens, unpublished). The diversity of food 
sources and habitats characteristic of agroforestry systems may confer a 
lower susceptibility to such 'biodiversity accidents' than is likely in less 
diverse agricultural contexts. Specifically, the presence of leaf and woody 
litter at the soil surface is likely to stimulate the activity of decompacting 
species such as ants and termites, with positive effects on water infiltration 
(Mando and Miedema, 1997). 

Conclusions and research needs 

Agroforestry systems are a promising alternative to the intensive 
monocropping systems which were promulgated during the 40 years or 
so of the green revolution. Soil macroin vertebrates, especially the group 
referred to as ecosystem engineers, are part of the biological resources that 
need to be managed in these systems. Invertebrates are not only indicators 
of the health of the system; they are also actors participating in the 
maintenance of soil physical properties, short-term recycling of nutrients 

Soil Macrofauna 317 

and longer-term protection of nutrients and organic matter in their 

The direct management of soil invertebrate communities, by, for 
example, inoculation with earthworms, is difficult and costly. This makes 
understanding the effects of plant communities and management practices 
on invertebrate communities very important, so that these can be 
optimized through appropriate design and management of agroforestry 
and other land-use systems (Senapati et al. , 1999; Lavelle and Spain, 2001). 
In addition to the maintenance of a favourable microclimate and the 
avoidance of tillage in at least part of the system, the greatest potential 
of agroforestry practices to benefit soil invertebrates lies in the provision 
of diverse and abundant organic resources. More research, however, 
is needed into ways of optimizing the supply of organic resources in 
different agroforestry practices and under various pedoclimatic conditions, 
with special attention given to the effects of litters of different quality, 
including mixtures of different litter types, on invertebrate communities. 
Generating information to reconcile the partly contradictory objectives, 
on the one hand, to provide sufficient assimilable organic matter (through 
rapid litter transformation) and, on the other hand, to maintain a 
continuous litter layer that protects the soil and serves as a habitat, requires 
an integrated research approach using experimental and modelling 

The composition of communities is also a relevant issue to address 
since the lack of functional diversity may have unexpected negative effects 
on soil function (Chauvel et al. , 1999). The composition of an invertebrate 
community depends on a suite of determinants that operate at different 
scales of time and space. Geographical constraints determine regionally 
or locally the pool of species present. Distance from a plot to reserves of 
species, such as primary or secondary forest, and the size and shape of the 
plot are significant primary determinants of the community composition. 
Inside the plot, the quality and quantity of litter produced by single plant 
components and the array of plant species present are of great significance. 
Sampling protocols and procedures for data analysis that are adapted to 
these sources of variability are necessary. They should integrate the spatial 
structure of the system at scales from landscape (watershed) to farming 
system and plot. 

The functional classification of invertebrates is another key issue to 
address. Functional classification systems are still in their infancy and rarely 
address the real effect of invertebrates on ecosystem function. 
Furthermore, no comprehensive classification for all soil invertebrates 
really exists. Future research should seek to interpret community 
compositions in terms of their effect on decomposition rates, organic 
matter dynamics over different time scales and effects on soil physical 

318 P. Lavelle etal. 

Finally, the energy balance of faunal activities has to be considered to 
design appropriate management options. These should be designed to 
provide enough energy to sustain an adequate level of faunal activity while 
allowing sufficient storage of recalcitrant or physically protected organic 
matter for the maintenance of exchange characteristics, storage of 
nutrients and stabilization of soil aggregates (Lavelle et at., 2001). 

1 6.2 Sampling of Macrofauna: the TSBF Methodology 

Sampling designs 

A method for the collection of soil macrofauna in the field, which had 
originally been proposed by Lavelle (1988), has been adapted to the needs 
of the TSBF (Anderson and Ingram, 1993) and has recently been re- 
evaluated and improved within the Macrofauna Programme of the 
International Biodiversity Observation Year (IBOY; D. Bignell et al., 
unpublished). The original sampling design comprised ten samples of 
25 cm X 25 cm and 30 cm depth, distributed every 5 m along a transect 
chosen on a random basis. A problem with this design has been the 
strongly aggregative distribution of most invertebrate groups in the field, 
which results in high variance and creates autocorrelation among the 
sampling points. To avoid this problem, it is now recommended that the 
distance between sampling points is increased, whenever possible, to at 
least 20 m. 

In agroforestry systems, two alternative protocols may be applied, 
which are stratified sampling or systematic sampling on a regular grid (see 
Section 3.4). With stratified sampling, samples of soil macrofauna are 
collected beneath trees of different species, if tree species effects are 
expected, and in the different individual components such as grass bands, 
crop strips or hedgerows. Ten samples per unit and certainly not fewer 
than five are recommended to detect significant differences between 

When considering communities at the scale of a farm or watershed, 
sampling on a regular grid is recommended. This design provides an 
average estimate of soil macrofauna communities in large landscape units 
covered by the grid and allows a stratification a posteriori by separating 
points on the basis of the local plant cover. Another advantage of this 
protocol is that the spatial structure of the plot is explicit and can be 
analysed with geostatistical methods (see Section 3.6). This is of great value 
in landscapes with land cover mosaics where movements of fauna between 
adjacent plots may determine the composition of communities. Spacing 
between sampling points will depend on the size of the area and the 
number of samples that can be processed with the available resources. 

Soil Macrofauna 319 


The sampling area is delimited, preferably with a metallic frame 
corresponding to the size of the sample (25 cm x 25 cm) with vertical blades 
of 5-10 cm on all sides that help to fix the device in position and to cut 
through the litter and topsoil. If litter is present, it is collected and kept in 
a hermetic plastic bag for further sorting. The soil sample is then isolated 
by cutting the soil vertically to a depth of 30 cm with a flat spade or 
machete. A ditch is dug on one of the four sides, and successive 10 cm 
strata are cut and kept in separate plastic bags for further processing. 

Hand-sorting is carried out on large flat trays (40 cm x 60 cm and 
5 cm deep). A handful of litter or soil at a time is spread in the tray to allow 
the capture of all individuals that can be seen with the naked eye. Controls 
of the accuracy of the method have shown that a variable proportion of 
individuals is found. This proportion may vary from 10% to almost 100% 
depending on the size, colour and mobility of the invertebrates (Lavelle, 
1978; Lavelle and Kohlmann, 1984). In spite of the relatively low and 
variable efficiency of the sorting, the method has proven robust and highly 
repeatable in a wide range of tropical and temperate areas with forest, 
crops or grass cover (Lavelle and Pashanasi, 1989; Dangerfield, 1990; 
Decaens et ai, 1994; Lavelle et al., 1994). The results are normally used 
directly in multivariate analyses for the comparison of different sites or 
treatments. Where absolute values are important, rather than relative 
differences in community structure between sites, correction factors may 
be established for different faunal groups at the beginning of a sampling 
programme. This is done by sorting samples both in the field and by heat 
extraction (Berlese funnels) for litter and washing-sieving extraction for 
mineral soil (see, for example, Lavelle and Kohlmann, 1984). This 
approach could be especially useful in long-term studies at a single site. 

It will normally take one person-day to cut and hand-sort two or three 
samples in the field. Since variance is always very high due to the 
inherently heterogeneous distribution of soil invertebrates, the 
recommended number of samples does not guarantee that statistically 
significant differences between treatments or sampling positions are 
obtained, even if effects are relatively large. 

Identification and data storage 

Although no precise guidelines on identification can be given at this stage, 
the invertebrates collected should be separated into a minimum list of 
major taxa for entry into a standardized database (Table 16.2; Fragoso et 
a!,., 1999). Whenever possible, more precise identifications should be used. 
Families will generally yield 30-60 entries. Identifications at the species 

320 P. Lavelle etal. 

Table 16.2. Minimum list of taxonomic units for soil macrofauna for inclusion in 
the MACROFAUNA database. 3 




Coleoptera adult 

Coleoptera larvae 

Total Coleoptera 




Total Myriapoda 


Dictyoptera (Blattoidea) 



Lepidoptera larvae 



Other macrofauna 

a After Fragoso etal. (1999). 

level should be attempted if the requisite taxonomic expertise is available. 
General identification keys are not yet available but identification tools will 
be progressively released as part of the IBOY programme on macrofauna 
described above. These tools will be made available along with access to 
the MACROFAUNA database on the Internet ( 
The use of morphospecies or recognizable taxonomic units (RTUs) is 
recommended when identification is not possible (Williams and Gaston, 
1994; Stork, 1995). 

Populations of each taxon are characterized by their numbers 
(individuals per m 2 ) and biomass. Biomass is preferably expressed as fresh 
weight of preserved invertebrates. Expressing biomass as dry weight 
considerably increases the amount of work and does not increase 
comparability of results unless guts are emptied of their soil or litter 

1 6.3 Other Sampling Methods 

Depending on the purpose of the study and the available manpower, 
alternative methods may be used such as the rapid biodiversity assessment 
(RBA) method (Oliver and Beattie, 1993; Jones and Eggleton, 2000), or 

Soil Macrofauna 321 

pitfall traps designed to catch the fauna that moves across the soil surface. 
Collection of biogenic structures deposited at the soil surface may be an 
acceptable surrogate to the evaluation of abundance, diversity and activity 
of invertebrate engineers. 

Pitfall traps 

Pitfall traps are pots (approximately 10 cm in diameter and 15 cm in 
depth) that are fitted into the soil in order to catch the vagile fauna. Ants 
and non-social litter arthropods (Arachnida, Coleoptera, Myriapoda) are 
the most commonly caught invertebrates; earthworms, termites and all 
truly endogeic soil fauna are less efficiently sampled or missed altogether. 
The following protocol was designed to sample surface arthropods in 1 
km 2 square plots in European landscapes as part of the BIOASSESS 
programme (J. Niemela, unpublished). Sixteen sampling points are 
regularly distributed across the area. At each point, four traps are installed 
at the four corners of a 5 m x 5 m square. Traps 8 cm in diameter are 
partly filled with either ethylene glycol or propylene glycol. They are 
covered with a rain cover placed a few centimetres above the trap. The 
trapping period extends over a minimum of 10 weeks and traps must be 
checked frequently to avoid decay and loss of material. Other sampling 
designs are needed for studying small-scale patterns within agroforestry 
plots, such as stratified sampling under different tree species, or sampling 
shaded and unshaded or mulched and unmulched areas. Traps are 
particularly suitable for such purposes (Lavelle and Kohlmann, 1984; 
Hanne, 2001). 

Pitfall traps may be made more efficient or selective by adding specific 
food substrates such as jam and canned tuna fish (ants), vertebrate dung 
or faeces (scarabeid beetles) or wood pieces (xylophagous termites). A 
different trap design specifically for termites consists of plastic tubes of 
about 5 cm diameter with lateral openings which are inserted in the soil 
so that the openings are below the soil surface. The tubes are covered with 
a lid. Pieces of wood are placed inside the tubes and are regularly 
controlled for the presence of termites (Su and Scheffrahn, 1996). 
Preliminary tests with different types of wood are recommended (Hanne, 

Collection of biogenic structures 

Soil ecosystem engineers produce a great variety of biostructures that they 
may deposit at the soil surface, hang on tree trunks or incorporate into 
the soil. Earthworms produce casts, middens, galleries and resting 

322 P. Lavelle etal. 

chambers. Earthworm casts are all made of faecal material that is generally 
richer in fine and organic particles than the bulk soil. Termites build 
epigeic nests of highly variable shapes and sizes, at the soil surface and/or 
in the soil, on tree branches or inside rotten trunks. They also dig a wide 
range of galleries and chambers and deposit surface covers and runways. 
Termite biofabrics and deposits may be made either of faecal material or 
material simply separated and worked with their mandibles. Non-faecal 
material may be either soil or carton, which is a soft mixture of excreta, 
undigested plant material and inorganic elements. The texture and 
organic contents of termite structures may be highly variable. Ants do not 
comprise geophagous populations and all their constructs are made with 
soil that has been removed with mandibles and deposited outside the 
endogeic nests, most often as accumulation of loose soil aggregates. In 
some cases, however, ants may build rather compact epigeic earthen nests, 
like Camponotus punctulatus in subtropical South America. 

Quantifying the abundance, diversity and composition of these 
structures may be considered as a functional substitute for the quantitative 
evaluation of soil macroin vertebrate communities (see Box 16.1). As such, 
this approach may be used to assess the impact of different agricultural 
practices. Sampling protocols need to be adapted to the local conditions. 
Large, long-lasting structures should be counted once or twice per year 
over large areas, whereas short-lived structures like most earthworm casts 
or ant deposits can be assessed by removing them from a surface and then 
measuring how many new structures are produced in a given period of 
time. This must be done in periods without rainfall, which may wash away 
the most fragile structures. Biostructures in the soil can be studied and 
quantified with soil micromorphological techniques (see Section 10.6). 

16.4 Manipulative Experiments 

The effect of invertebrates on soil processes can also be assessed with 
manipulative experiments in the field. Litterbags are widely used to assess 
decomposition of litter under field conditions (see Section 6.3 and 
Anderson and Ingram, 1993). Litter is enclosed in a mesh made of an 
undecomposable material and left on or in the soil under field climatic 
conditions and exposed to the community of decomposers. The 
participation of invertebrates in decomposition is classically evaluated by 
using different mesh sizes that permit access by invertebrates of different 
sizes (e.g. 0.2, 2 and 5 mm). This technique has been widely used, so that 
it produces data that are readily comparable with other data sets. The 
major drawbacks are the confinement of organisms due to the effect of the 
mesh, an altered microclimate in the mesh bag and an imperfect contact 

Soil Macrofauna 323 

with underlying litter or soil layers. For a detailed discussion of the method 
see Section 6.3. 

Effects of soil fauna and other biological and non-biological agents 
influencing the soil structure, such as roots or wetting-drying cycles, can 
be assessed by exposing undisturbed soil monoliths in a given soil 
environment and monitoring changes in their physical structure and other 
properties (Blanchart, 1992; Barros et at., 2001). The formation of large 
aggregates from a soil previously sieved at 2 mm under the influence of 
earthworms has also been studied with this method. Re-aggregation of soil 
by large compacting earthworms was quite fast, with the original level of 
aggregation being obtained after only 1 year (Blanchart, 1992). 

Chapter 1 7 
Soil Erosion 

M.A. McDonald, 1 A. Lawrence 2 and P.K. Shrestha 3 

1 School of Agricultural and Forest Sciences, University of Wales, 
Bangor, Gwynedd LL57 2UW, UK; 2 Environmental Change Institute, 
University of Oxford, 5 South Parks Road, Oxford OXI 3UB, UK; 
3 Local Initiatives for Biodiversity, Research and Development 
(LI-BIRD), PO Box 324, Bastolathar, Mahendrapool, Pokhara, Nepal 

MA Synopsis 

There are two basic approaches to the study of erosion rates. In the first, 
sediment transport rates are monitored at the outlet from a catchment. 
Such measurements are relatively easy to make and they integrate the 
effects of erosion over a variety of land uses but there are problems in 
determining the source of the sediment within the catchment. The second 
approach involves measuring processes at a number of sampling sites 
within a catchment. The techniques vary with the processes involved. Such 
data are more difficult to collect but have obvious advantages in providing 
information about the spatial distribution of erosion rates within a 
catchment and the relative contribution of different land uses. The latter 
approach is the most applicable to agroforestry research at a field scale, 
seeking to investigate the effects of incorporating a tree component into a 
farmed environment. If both approaches are combined within the same 
catchment, the results of the former may serve as a test for the successful 
scaling up of the results of the latter. 

The need to evaluate rates of soil erosion in agroforestry arises from 
investigations on sloping land where soil transport is a major issue. The 
presence of trees in hillside landscapes complicates processes that affect 
soil erosion so that methodologies commonly used to evaluate it require 
adaptation for use in agroforestry, and it is important to have a precise 
definition of the objectives of the research. Considerable interest in 
agroforestry research has focused on the potential for erosion control, and 
the role of trees in the redistribution of soil and associated nutrients and 
moisture on farmed slopes. 

© CAB International 2003 . Trees, Crops and Soil Fertility (eels C. Schroth and 325 

F.L. Sinclair) 

326 M.A. McDonald ef al. 

Trees in agroforestry systems can affect rates of erosion in several ways: 
(i) they can act as a semipermeable barrier which slows the velocity of 
surface runoff, and hence its erosivity; (ii) they can increase infiltration by 
the same mechanism and also through the improvement of soil structure 
(see Chapter 10) and hence reduce runoff volume (which reduces nutrient 
losses in runoff but can increase nutrient leaching; Ramakrishnan, 1990); 
(iii) they can protect the soil from the impact of rainfall by production of 
a litter layer and/or prunings that are used as a mulch; and (iv) they can 
act as a sieve barrier to eroding soil and hence change the particle size 
composition of eroded sediments. 

The presence of a tree canopy has contradictory effects on erosion. 
On the one hand, trees reduce the quantity of water that reaches the soil 
through crown interception (see Section 11.1). On the other hand, the 
erosivity of rainfall may increase because the size and spatial distribution 
of raindrops are modified during crown passage (Armstrong and Mitchell, 
1988). Raindrops intercepted by tree leaves may coalesce on leaf surfaces 
before emerging as larger throughfall droplets, which are more erosive, 
beneath the tree crown. The distribution of drop sizes under trees with 
different leaf morphologies has been found to vary significantly, with 
larger leaves generally producing larger drop sizes (Calder, 2001), 
although drop size also depends on leaf surface characteristics and leaf 
inclination. The proportion of raindrops that are modified as opposed to 
passing through the canopy unaffected depends on crown cover and 
density. Tree height may also increase the kinetic energy and hence the 
erosivity of drops reaching the soil. Drops reach 95% of their terminal 
velocity in a free fall of 8 m (Young, 1997), so a further increase in height 
does not greatly increase their velocity when reaching the soil. Increased 
erosivity of throughfall over rainfall in agroforestry systems is often well 
understood by farmers (Thapa et al, 1995) and has been well documented 
scientifically (Wiersum, 1984). Some trees also produce high stemflow 
rates, which may increase erosion under certain conditions (Wiersum, 
1985). In light of this, it is unsurprising that experimental work has clearly 
shown that it is not the presence of the tree canopies, but rather the direct 
soil protection offered by a continuous litter layer that is responsible for 
low erosion rates in undisturbed forest (Wiersum, 1985). This has clear 
consequences for erosion control in agroforestry systems, where the litter 
layer may be disturbed during crop cultivation. 

There are two principal ways of integrating trees and shrubs to control 
erosion in agroforestry, which differ in the relative importance of barrier 
as opposed to soil cover effects. These are: 

• dense, mixed agroforestry systems (multistrata systems and perennial 
tree crop combinations); and 

• spatially zoned systems (contour hedgerows and other systems of trees 
and shrubs in contour-aligned belts). 

Soil Erosion 327 

The former control erosion by means of a dense, surface ground cover 
of plants, litter or mulch, which, if maintained, is clearly effective 
(Wiersum, 1984; Young, 1997), though few experimental studies have 
been performed. Spatially zoned systems have received greater attention, 
and experimental results have been instructive in establishing the role of 
trees on slopes in increasing spatial heterogeneity in a number of soil 
properties (see, for example, the discussion of terrace formation below). 
In an agroforestry context, a fundamental question is how landscapes with 
trees are positioned in a continuum from pure cropland to natural forest 
(van Noordwijk et al, 1998) and it is important that the processes 
elucidated in studies of trees at various densities and in various spatial 
arrangements are considered in their landscape context. 

Approaches to measuring soil erosion in agroforestry research 

Understanding the interaction of trees and slopes on the redistribution of 
soil and nutrients within the area being evaluated is of fundamental 
importance. Strong evidence has been found for a scouring effect in 
contour hedgerow systems on hill slopes whereby soil is lost from the upper 
part of the alley between hedgerows and accumulates in its lower part, 
leading to a transfer of surface soil properties between the two parts of the 
alley (Garrity, 1996). Usually, the effects of this redistribution of soil within 
the alleys is such that, after some time, the differences in soil properties 
between the upslope and the downslope side of an alley are greater than 
those between areas with and without hedgerows (Garrity, 1996; Agus et 
al, 1997; Poudel et al, 1999; McDonald et al, 2002). The magnitude of 
this redistribution depends on the intensity of tillage (Garrity, 1996) and 
the hedgerow spacing. The choice of hedgerow spacing depends on 
steepness of a slope and, in turn, determines the ratio of the area of land 
that is between hedgerows to that which is under hedgerows, and hence 
the available land for cultivation. This ratio is also important because 
erosion control processes, such as high infiltration rates of water, occur 
predominantly under the hedgerow (Kiepe, 1995), which may have 
significant implications for nutrient losses by leaching (see Chapter 7). 

The significance of the spatial variability in soil properties for 
experimental design was discussed by Mueller-Harvey et al. (1989) and 
highlights the crucial importance of including a sufficient number of 
independent experimental blocks, which control for variation in both site 
conditions and farmer practice, in order to produce conclusions of general 
applicability. If there is significant between-block variation, treatment 
effects will not be detected unless it is removed from the residual sum of 
squares (see McDonald et al, 2002, for an example). Location and layout 
of blocks should also take into account the evidence provided by Pickup 

328 M.A. McDonald ef al. 

(1985), Westoby (1987), NortdiS etal. (1990), Busaccarf al. (1993), Garrity 
(1996) and van Noordwijk et al. (1998) that on hillslopes there will be major 
transfers of nutrients within the slope between areas of net erosion 
(sources) and net deposition (sinks) depending on the topography of the 
slope (even fine-scale changes that are undetectable to the human eye). 
Mid-slope sites may undergo little or no net soil loss. A major impact of 
changes in land use, such as incorporation of trees, is likely to be on this 
rate of within-slope transfer. Farmers have learned to adjust their land- 
use practices to accommodate the effects of erosion by, for example, 
planting different crops in zones of net loss and net deposition (van 
Noordwijk et al, 1998), so that expensive technologies that minimize such 
transfer by surface runoff and erosion may not actually address farmers' 
needs. Ramakrishnan (1990) was also concerned that, on long hillslopes, 
minimization of surface runoff, by increasing infiltration, may not be 
advantageous to overall soil fertility and agricultural productivity, because 
of increased nutrient losses through leaching. Therefore, the objectives of 
studies need to be carefully framed to distinguish net impacts of changes 
in land use at the individual plot scale, at the whole hillslope scale, or on 
the rate of discharge into rivers at the bottom of the hillslope. 

McDonald et al. (2002) observed that, for 14 measured soil properties, 
variation between blocks was much greater than that between agriculture 
and agroforestry treatments in an experiment in the Blue Mountains of 
Jamaica. This indicated that, for individual farmers operating small-scale 
slash-and-burn agriculture and seeking to maximize fertility of their 
farmland, selection of more fertile sites in which to establish new farmland 
was a more important decision than whether or not to adopt agroforestry 
practices, at least over the 5-year time course monitored by the 
experiment. Forest restoration through secondary succession during a 
fallow period may be the most effective mechanism for the recovery of soil 
fertility in sloping lands of the humid tropics (McDonald and Healey, 
2000). However, adoption of agroforestry practices may become more 

• if shortage of farmland and restrictions on tenure prevent farmers 
from selecting sufficiently fertile sites for new farm establishment; 

• if long periods of cultivation lead to soil fertility falling to very low 
levels; or 

• if it is the other products and services provided by agroforestry 
practices that are important rather than their impact on soil fertility. 

Conducting experiments over a time frame sufficient to indicate long-term 
trends and to adequately account for the heterogeneity of the environment 
is essential in this type of research. 

Accumulation of soil material above the trees is commonly observed; 
hedgerows clearly do have a sieve-barrier effect that causes changes in soil 

Soil Erosion 329 

texture over time. The higher sand concentrations above trees may partly 
be because sand is more easily moved down-slope than clay, which 
commonly leads to higher clay contents on sloping land (Fielding and 
Sherchan, 1999), and also reflects the fact that the hedgerow sieves trap 
larger sand particles more effectively than the finer silt and clay ones. 
The accumulation of soil above the hedgerows progressively reduces the 
slope angle, resulting in the formation of terraces, and, therefore, reduces 
runoff generation. Higher potassium and sodium concentrations under 
hedgerows may indicate that transport above the soil surface is a significant 
mechanism in their movement, either in the eroded soil that accumulates 
there or in runoff water, which is likely to have much higher rates of 
infiltration under the hedgerows (Kiepe, 1995). Ongprasert and 
Turkelboom (1996) indicated that on hillslopes in northern Thailand, 
higher soil fertility is found under contour hedgerows than between them. 
Such results are influenced by the relative importance of the sieve-barrier 
effect, which tends to concentrate fertile topsoil under the hedgerows, the 
transfer of prunings to the between-hedgerow zone, which tends to 
redistribute the fertility (Young, 1997), and the removal of nutrients from 
the between-hedgerow zone with crop harvests. While trees also influence 
soil fertility through litterfall inputs and root uptake, the spatial 
distribution of these two processes is highly variable. 

Studies of erosion within the context of agroforestry research must 
incorporate an additional spatial dimension. All farming systems operate 
within the socioeconomic and natural environment, but the forces of 
gravity acting upon transport processes on sloping land create complex 
patterns of spatial variability (Gardner, 1997). Simply measuring rates of 
soil movement will not be adequate for understanding the complexity of 
factors occurring when trees are introduced in hill environments. 

Developing farmers' criteria for evaluating soil erosion 

Increasingly, development-oriented research aims to be participatory, and 
collaborations between scientists and farmers have been particularly 
significant in the field of soil erosion. Studies of farmers' knowledge about 
soils, erosion and fertility are still much more common than action research 
which seeks ways in which farmers can address soil erosion through 
developing their own knowledge and actions. 

Such academic studies are important starting points for recognizing 
the strengths of farmers' knowledge, and where it differs from scientific 
knowledge. Results tend to emphasize the detail of farmers' knowledge, 
recognizing a wide range and diversity of soils within their resource use, 
and the utilitarianism of their knowledge (Woodgate, 1994; Zimmerer, 
1994; Carter and Murwira, 1995; Lamers and Feil, 1995). Similarly, 

330 M.A. McDonald ef al. 

farmers' knowledge about soils and soil-related processes has been found 
to be quite consistent across farming communities, especially with similar 
agro-climatic conditions. A study in Nepal (P.K. Shrestha, unpublished) 
shows consistency in farmers' knowledge about soil physical properties 
and their association with soil erosion and soil fertility across three different 
communities in the western middle hills. Some of such knowledge, 
however, is patchy or specific to a location, community or even individual 
farmers within a community due to differences in agro-climatic conditions, 
cultural tradition, gender and personal experience. 

Farmers' knowledge of soils and soil-related processes has also been 
found to be complementary to scientists' knowledge and this can help to 
focus further research efforts. For example, in Nepal, comparison of 
farmers' perceptions of the timing of soil loss and runoff with actual 
measured soil loss and runoff shows that they are strongly correlated, and 
this may provide a shortcut to research monitoring. However, in the same 
areas it was found that farmers do not know about leaching losses (Shrestha 
et al., 2003), and it has been found elsewhere that farmers know less about 
below-ground processes, which are difficult for them to observe, than 
about above-ground processes (Sinclair and Walker, 1999). Such 
knowledge gaps represent problems that farmers are not likely to be able 
to address by themselves, and can help to identify topics suitable for 
participatory research. 

There are several reasons why it is important to involve farmers in soil 
erosion research, either through conducting trials on their land, inviting 
them to observe trials on research stations, or supporting them in 
conducting their own research (e.g. Okali et al., 1995). The benefits are as 

• The research will be better adapted to the environment: farmers have 
detailed local knowledge of what is often a highly heterogeneous 
environment, especially in hilly or mountainous areas where soil 
erosion studies are likely to be carried out. For example, studies of 
farmers' knowledge in Nepal showed that agroforestry systems have 
quite different potentials for soil conservation in different ecological 
situations in the middle hills. At lower elevations, trees on cropland 
are acceptable to farmers, whereas at higher altitudes and on north- 
facing slopes, farmers find the competition between trees and crops 
too severe (P.K. Shrestha, unpublished). This finding can help to guide 
appropriate technology design. 

• Explanations for results may be better informed by farmers' own 
observations of change over many years. There are numerous 
documented cases where historical change known to farmers has 
brought about, or alleviated, the problems visible at present. Preston 
et al. (1997), for example, have shown that contemporary severe soil 

Soil Erosion 331 

erosion in Bolivia is most probably due to overgrazing in the past, and 
that vegetation is in fact recovering at the present time. 

• If farmers are able to observe experiments, participate in the 
measurements or conduct their own research, they are more likely 
to be convinced by the results and to interpret them in the context 
of their livelihood systems so that the results can be incorporated 
into their farming methods in appropriate ways. A tradition of 
participatory technology development has arisen among non- 
governmental organizations and merged with farming systems 
research methods (Martin and Sherington, 1997), but these are 
particularly challenging in the context of soil conservation research, 
and new methods are still being developed (Lawrence, 1999; Lawrence 
etal, 2000). 

• Participatory approaches can lead to shared knowledge and enhanced 
awareness within the community, and directly to farmers identifying 
indicators to measure soil erosion in their own systems, and to monitor 
changes in soil erosion as they or others conduct research into the 
effectiveness of interventions. 

In order for farmers to participate in such experiments, they must use 
methods that are accessible and meaningful to them. Attention has focused 
in particular on seeking indicators of soil erosion which are identified by 
farmers and which can be measured by them. The relatively new group 
of methodologies which have grown up around participatory monitoring 
and evaluation (Estrella et al., 2000) are helpful here, and collective 
experience suggests that farmers' indicators are likely to be visual, 
observable or countable but not requiring complex measurements or 

Two case studies of indicator development summarized here contrast 
the priorities and approaches of farmers in systems where they see soil 
erosion as a low-priority problem (Andean foothills, Bolivia) and a high- 
priority problem (middle hills, Nepal). In both cases, farmers and scientists 
collaborated to monitor and evaluate the success of agroforestry and other 
interventions for controlling soil erosion on sloping cultivated land. In 
Bolivia, the communities involved are living in semiarid mountainous 
areas with road access to the city, resulting in migration and relatively low 
population densities in the rural communities. Because of this and the 
lower rainfall, free-range livestock are also an important component of the 
farming system. In contrast, in the research areas in Nepal, population 
densities are much higher, annual precipitation is much higher (but, as in 
the Bolivian case, strongly seasonal and erosive) and land use is much more 
intensive. As the following discussion shows, the farmers in Bolivia were 
initially not very interested in the soil erosion aspects of the trials, but 
recognized the value of this aspect with time. 

332 M.A. McDonald ef al. 

In the Bolivian case (Lawrence et a!,., 2000), scientists' explanations of 
the value of contour hedgerows based on erosion control met with 
scepticism among farmers, who were, however, interested in the possibility 
of increasing fodder production, particularly during the dry season. As 
the trials developed, farmers changed the criteria that they were using to 
assess the trials. At an early stage, when the contour hedgerows were still 
small, farmers focused on potential losses brought about through the trials, 
and highlighted the increased labour required to prepare the land for 
sowing when stubble cannot be burnt; risk of losing hedgerows through 
browsing; and compatibility with farming practices such as ploughing by 
oxen. However, after one growing season, participating farmers and their 
neighbours began to notice both soil and moisture retention around the 
hedgerows, and began to express interest in establishing further trials. At 
this stage they were invited to explicitly identify indicators to evaluate such 
trials, and included soil colour, soil cover, humidity and soil content of 
runoff in their indicators, showing a change in perception. However, 
increased crop and animal production were still their priorities. 

In the more structured research context in the middle hills of Nepal, 
where farmers were specifically invited to monitor soil erosion and had 
more incentive to do so, given higher population densities and more 
intensive land use, farmers identified the following much more detailed 
and specific set of indicators of soil erosion (Shrestha et al. , 2003): changes 
in the number of stones exposed on the surface; exposure of the base of 
terrace risers; changes in the height of the terrace risers; formation of rills 
or gullies; changes in soil depth; exposure of crop and tree roots; change 
in plant vigour and health; change in crop yield; change in outward slope 
of terraces; turbidity of runoff water; change in soil colour; change in soil 
structure. By explicitly developing soil erosion indicators with farmers, the 
results in Nepal provide a more detailed set of measures to analyse the 
success of soil conservation measures, and improve the chances of farmers 
and scientists communicating well. 

There are several lessons to be drawn from this experience. The 
Bolivian experience shows us that farmers' evaluation criteria are context- 
specific and may evolve during the course of a trial; they may also vary 
according to the season. The indicators adopted by farmers are visible and 
readily observable and, in both cases, farmers judge success according to 
their own direct experience, which often relates specifically to their own 
fields. The work in Nepal suggests that, although farmers look at what 
their neighbours are doing, unless they are actually working the soil and 
directly observing the colour, texture and runoff quality, they will have 
only a vague idea of the differences in soil erosion between plots, and 
between different agroforestry interventions. 

A trial is rarely evaluated solely on the basis of its control of erosion. 
Farmers are interested in looking at the effect of agroforestry (or other 

Soil Erosion 333 

technologies) on their whole livelihood system, including labour 
availability, food security and resource flows for priority enterprises such 
as fodder for cattle. As Bunch (1999) has pointed out, rapid returns and 
economic benefits are more important to poor farmers with a short 
resource management horizon than soil erosion per se (see also Chapter 
2). This last point is an important one, and not unique to the case studies 
described here. It is linked to a more serious point that scientists must be 
careful not to ignore: that, even where soil erosion appears to the outside 
observer to be severe, it may not be perceived as the principal problem by 
farmers themselves. There are plenty of case studies where soil erosion is 
identified as a constraint to productivity, particularly in semiarid systems, 
but there are others where outsiders have focused on this as a researchable 
constraint which can be addressed by a technological fix, whereas local 
residents may be aware of other priorities or sociopolitical reasons why 
such technological fixes will not work (Blaikie and Brookfield, 1987; 
Preston et al., 1997). This is why soil erosion research must take place 
within a social research context. 

The indicators that were developed with farmers in the above case 
studies were based on an explicit process facilitated by researchers. 
Farmers do not tend to consciously develop indicators, and it is sometimes 
the case that participatory monitoring and evaluation can place 
unwarranted demands on farmers, if the results are only of use to the 
researchers. Nevertheless, by participating in experiments which have a 
short-term incentive, such as fodder production, but which at the same 
time may reduce soil erosion, farmers may change their perceptions: those 
in the Bolivian case commented that they had always thought soil erosion 
was an unstoppable, 'natural process'. Within 2 years of beginning contour 
hedgerow trials on some farmers' fields, others were copying them 
spontaneously, not only for fodder production but also because soil erosion 
was reduced and, most importantly in these semiarid hillsides, because 
water was retained in the soil. 

It is therefore important to pay close attention to the process that is 
used to develop indicators. The same core process was used in the two 
cases described here (Table 17.1). Farmers are often unfamiliar with the 
concept of formal evaluation using indicators, but they are constantly 
observing change in the environment around them, and will be able to 
discuss good and bad changes easily. With careful facilitation the 
researchers will be able to list key words that show important change and 
discuss a list of them with farmers. For example, in Nepal, farmers 
responding to trials designed and managed by scientists noted their 
perceptions of good and bad aspects of the trials. These included 
observations about labour requirements, soil deposition and changes in 
crop yield, and provided a starting point for developing indicators. Once 
the indicators have been used in explicit evaluation and comparison of 

334 M.A. McDonald ef al. 

Table 17.1. The process of developing indicators with farmers. 

Situation analysis, to improve understanding of the interlinkages in farming 

systems and to identify farmer perceptions 

Discussions with farmers regarding individual experiences with systems 

changes as a result of incorporating new farming practices 

Discussions with farmers (individually or collectively) regarding their 

expectations of incorporating new farming practices 

Indicator development together with farmers and researchers to show 

farming systems changes and impacts of new technologies 

Refining of systems indicators in the field by involving more farmers 

Use of matrix diagrams based on the identified indicators to rank and score 


interventions or plots, their utility will be much more apparent to the 
participating farmers. 

Research priorities 

Haifa century of failed soil and water conservation projects in tropical 
developing countries demands a reconsideration of strategy (Critchley et 
al., 1994). Recognition of this and the changing attitude of soil and water 
conservation programmes since colonial times has placed much greater 
emphasis on people's participation in all aspects of project design and 
implementation (McDonald and Brown, 2000). This has resulted in the 
adoption, in some areas at least, of a more holistic attitude towards the 
maintenance of land productivity (e.g. Shaxson et al. , 1989; Douglas, 1994; 
Scoones et al., 1996). Much emphasis on helping farmers to improve land 
husbandry and less on efforts to combat erosion alone are expected to 
provide a more effective solution to an old problem. It has been postulated 
that, if farmers find it feasible and worthwhile to improve the structure, 
organic matter content, porosity and nutrient levels of the soil, natural 
fertility will be raised and soils will recuperate and runoff and erosion be 
diminished (Shaxson et al. , 1997). 

It is now well recognized that the impacts of erosion are experienced 
differentially, and that some land managers benefit from the processes of 
soil erosion and runoff through redistribution of water and nutrients on 
their fields; for example, Stocking (1996) cites examples from Kenya and 
Sri Lanka, and Rakotomanana (1991) from Madagascar. Other findings 
suggest that rapid soil erosion may be a passing phase and that when 
appropriate institutional, economic and social factors are in place then 
higher rural population densities may be associated with investment in 
environmental enrichment (Tiffen et al., 1994). The dynamics of change 

Soil Erosion 335 

Table 17.2. Considerations in defining research objectives. 

• Who is expected to use the research results? 

• Is it necessary to convince policy makers? 

• How heterogeneous is the area where research is to be conducted? 

• How heterogeneous is the farming community and who makes decisions 
about land use/who is affected by land use? 

• How are people (researchers and farmers) used to interacting in the culture 
in question? 

• How has land use changed historically - can farmers avoid wasting time 
addressing causes which are historical? 

and the interactions between these factors demonstrate that the 
assumption that more people inevitably means more erosion and greater 
poverty is not only simplistic but also sometimes misplaced (Blaikie, 1986; 
Blaikie and Brookfield, 1987). Crisis narratives have developed as a result 
of these simplistic assumptions leading to large-scale inappropriate and 
largely unsuccessful policy prescriptions, particularly in sub-Saharan Africa 
(Tiffen et al, 1994; Hoben, 1995; Rocheleau et al, 1995; Roe, 1995; 
Benjaminsen, 1998) but also in Himalayan regions (Ives and Masserli, 
1989). The emphasis in monitoring erosion in agroforestry should not be 
restricted to monitoring soil movement but should take a holistic approach 
to monitoring and appraisal, combining a variety of indicators appropriate 
to the objectives of the research. The most valuable approach may often 
be to use a combination of farmers' and scientists' criteria, but the research 
must be tailored in each case to certain considerations (Table 17.2). 

It should be recognized that sources of error are considerable, 
depending on the sampling procedures, and that absolute values of soil 
erosion may not be meaningful. However, relative measures of soil 
changes, which take account of spatial patterns of relative loss and 
accumulation and if monitored in a participatory manner with farmers 
actively involved in the design and implementation of the research, can 
provide reasonable confidence in evaluating agroforestry innovations. 

Therefore, this chapter highlights some of the most appropriate 
methods for agroforestry research and the special methodological 
considerations required when trees are involved in erosion control rather 
than describing standard methods of measuring erosion rates that are well 
covered in many useful syntheses (e.g. Hornung, 1990; Lai, 1994; Hudson, 
1995; Morgan, 1995; Stocking and Murnaghan, 2001). 

336 M.A. McDonald ef al. 

1 7.2 Quantitative Methods 

Runoff plots 

There is debate as to the value of using plot-based measurements in 
erosion research, as they isolate the plot from the upper hillside, and can 
interfere with lateral drainage patterns. However, they provide a realistic 
means of evaluating relative treatment effects in multivariate studies 
(Young, 1997; van Noordwijk et al., 1998). Historically, they are the most 
commonly used technique in soil erosion research as they measure runoff 
and soil loss directly. Runoff plots can be inexpensive and easily 
constructed, and thus a sufficient number can be installed to obtain a 
representative sampling of the major characteristics of an area such as 
slope, soil type, plant cover and agricultural practice. Morgan (1995) 
describes their construction and monitoring. Plots are rectangular with 
the long axis oriented up-slope, and consist of a border, an element to 
concentrate runoff at the bottom end, and a collector to contain the runoff 
and sediment produced. Plot borders may be sheet metal, wooden planks, 
concrete or earth bunds. The smaller the plot, the sharper the edges need 
to be to minimize the edge effect. Standard runoff plots are 22 m long and 
1.8 m wide, but plots containing trees should be much larger, not less than 
40 m x 10 m (Young, 1997) to counter the problems of tree roots spreading 
from one treatment plot into other adjacent plots (see Section 3.2). 
Collectors are generally a flattened funnel with a hinged cover to facilitate 
cleaning. The sheet metal floor of the element is installed at least 5 cm 
below the surface, with a strip of sheet metal on the up-slope side flush 
with the soil surface to provide a sharp boundary at the plot end, which 
empties into collecting drums with a system of divisors to cope with large 
volumes of runoff. The soil loss is measured by stirring the tank, taking a 
sample of suspended sediment, filtering and drying, and then calculating 
the total loss from knowledge of the tank volume. Since most runoff plots 
in developing countries are simple, low-cost installations, there are many 
sources of error depending on the sampling procedure. Zobisch et al. 
(1996) compared the manual sampling accuracy of field staff in charge of 
erosion plots in Kenya. The measurements displayed acceptable accuracy 
for measurement of runoff volumes, but the accuracy for soil loss 
measurements varied considerably, with an average sampling error of 
41.3%. Stocking (1996) attributes this error to inadequate mixing of the 
solution in the tank. Even with vigorous stirring, it is difficult to get a 
representative sample as clays are over- and sands under-represented in 
the sludge samples. For large tanks, a good approach is to empty them 
successively, e.g. with a bucket, and collect a suspension sample at regular 
intervals, e.g. from each fifth or tenth bucket, depending on the total 
volume. After emptying the tank, the sandy sediment on the bottom of the 

Soil Erosion 337 

tank is collected as a separate sample. After drying all samples, the quantity 
of eroded soil is calculated by multiplying the average concentration of 
suspended soil per unit volume of water from the suspension samples with 
the total water volume in the tank and adding to this the bottom sediment 
(T. Morshauser, unpublished). 

Runoff plots have been used in several agroforestry studies (e.g. 
Hurni and Nuntapong, 1983; Maass et al, 1988; Lai, 1989; Agustin and 
Nortcliff, 1994; Alegre and Rao, 1996) and modifications have included 
use of Gerlach troughs (Maass et al, 1988; McDonald et al, 2002) or 
catchpits (Hellin and Larrea, 1997). These are described below. 

Difficulties arise in scaling up results from such plots to farmers' fields, 
hillsides or catchments, so, for a better understanding of erosion as a 
multiscale process, other indicators of processes within standard plots 
should also be observed (see subsequent sections). This is particularly 
important as agroforestry research evolves from a focus on field-scale 
technologies to a more complete consideration of landscape-scale processes 
and constraints involving a farmer-led process of technology development 
(van Noordwijk et al., 1998). 

Back-sloping terraces: a special case 

Runoff plots were developed for uniform slopes with gentle gradients in 
the USA (Wischmeier and Smith, 1978). Hillside terraces are far smaller 
than fields, and a standard runoff plot will therefore encompass upwards 
of three terraces. As yet, there is no standard means of quantifying runoff 
and erosion losses from such terrain (Hudson, 1992). With the back- 
sloping terraces common throughout South-east Asia, bounded plots will 
dissect lateral drainage paths (toe-drains), and the collecting drums may 
effectively collect only from the lowest terrace, giving artificially low values 
(Bruijnzeel and Critchley, 1996). Bruijnzeel and Critchley (1996) have 
suggested the concept of a natural boundary erosion plot based on a single- 
terrace unit (bed and up-slope riser) with its actual boundaries (Fig. 17.1). 
The toe-drain is located at the foot of the riser, so runoff and eroded 
material are channelled out of the individual terrace unit and can be 
conveniently monitored at the outlet. This will also collect uncontrolled 
surface run-on or subsurface water input from upper slopes (Gardner et 
al, 2000), so such measurement devices should be installed on several 
terraces in different positions within the slope to obtain an overall picture. 
Bruijnzeel and Critchley (1996) suggest another, specific metholodogy to 
determine the source of the eroded material. This involves installing a set 
of removable terrace riser troughs, which collect runoff, sediment washed 
down from the riser face and, presumably, any subsurface water flow 
emerging from the riser face (Fig. 17.1). 


M.A. McDonald et al. 

Natural boundary 
erosion plot 


Outlet point 
sits for V notch 

Fig. 17.1. Natural boundary erosion plot with one terrace riser trough (TRT) in 
place (reproduced with permission from Bruijnzeel and Critchley, 1996). 

This area of research will become more relevant to agroforestry 
research if the objective is to control riser erosion and increase the 
productivity of terraced systems, e.g. by inclusion of perennial fodder 
grasses and legumes, with stabilizing fodder trees planted on the top of 
the riser (Shrestha et al, 2003). 

Cerlach troughs 

This method is described in detail by Morgan (1995). Gerlach troughs are 
simple metal gutters, usually 0.5 m long and 0.1 m broad, closed at the 
sides and fitted with a removable lid. An outlet pipe runs from the base of 
the gutter into a covered collecting vessel (Fig. 17.2). They are normally 
sited at different slope lengths and the collecting area assumed to be equal 
to the width of the trough times the slope length (Fig. 17.2). On non- 
uniform slopes, the gutters should be sited to collect down-slope and not 
lateral drainage. The lack of plot boundaries eliminates any edge effects. 
These gutters can therefore be used without runoff plots (e.g. McGregor, 
1988) to investigate the interaction of soil status and soil erosion with land 
use. They provide a relatively efficient means of covering a range of 
conditions within a site. They have been used to collect runoff and eroded 
soil from small bounded subplots nested within a treatment plot (e.g. Ross 
et al., 1990) but, clearly, this is not an applicable approach in plots 
containing trees. Small plots will allow investigations into infiltration and 
the effects of rain splash but are too short for studies of overland flow 
except as a transporting agent for splashed particles (Morgan, 1995). 

Soil Erosion 


u Gerlach trough 

Siting Gerlach troughs on slope profile 

10 cm 

Hinged lid 


Gerlach trough 

Runoff and 

Ground slope 




Fig. 17.2. Gerlach troughs (reproduced with permission from Morgan, 1995). 


Catchpits are collecting pits that are dug into the ground at the bottom of 
the study area and lined with plastic or concrete to render them 
impermeable and that fill with runoff water and eroded soil. Catchpits can 
be effective in measuring amounts of sediment lost and runoff volumes if 

340 M.A. McDonald ef al. 

the reservoir is of sufficient capacity, although smaller pits that only catch 
an unknown proportion of the sediment can still be used to obtain 
comparative information. Pits of around 10 m 3 have been used in Thailand 
(Hudson, 1995). Obviously, durable plastic is required that will not be 
holed by soil fauna or degraded by sunlight. Hellin and Larrea (1997) used 
catchpits lined with plastic that had been punctured with small holes so 
that runoff water slowly dissipated and the sediment load was retained 
until end-of-season measurement. They found soil loss to be within the 
same range as that recorded by barrels below similar plots. Catchpits have 
the added advantage of providing a visual display for farmers of the 
effectiveness, or not, of the practice under test. 

Stakes and pins 

Repeated measurement of the height of the ground surface at stakes or 
erosion pins has been used to estimate erosion. Instrumentation consists 
of a long nail or stake and a large washer. Stakes should be made from 
metal or another resistant material as wooden stakes are rapidly attacked 
by termites under tropical conditions. At the time of installation, the nail 
is driven into the soil, with the washer lying on the soil surface. The 
distance from the head of the nail to the top of the washer is then measured 
(Fig. 17.3a). Erosion moves material from around and beneath the washer. 
Remeasurement of the distance between the top of the nail and the top of 
the washer provides a measure of the rate of erosion in the intervening 
period (Fig. 17.3b). If the washer has protected the soil from raindrop 
impact, so that it now stands on a pedestal, the pedestal must be removed 
before measurement, so that the washer lies at the general ground surface. 
The advantage of using the washer is that it gives a firm surface from which 
to measure. The washer can also be removed between two measurements. 
Marking the pins with bright paint aids relocation. Such measures are 
cheap and easy to install over a wide range of conditions. A disadvantage 
of the method is that it cannot be used on tilled soil because of the change 
of soil surface caused by tillage and the subsequent resettling of the soil. 

Instead of measuring soil removal by erosion, pins can also be used to 
measure soil deposition. In a sloping area in central Togo, G. Schroth 
(unpublished) used the method to determine patterns of sediment 
deposition in strips of trees and hedgerows which had been planted on 
the contours to collect soil lost from adjacent tilled crop fields where the 
pin method could not be used. Pins could be a cost-efficient method to 
determine patterns of net erosion and deposition over entire slopes (see 
Section 17.1). 

Instead of measuring directly around the pins, a cable or other rigid 
device can be placed between two pins or stakes. With this improvement 

Soil Erosion 


(a) Washer 



Fig. 17.3. Measurement of erosion and deposition at stakes: (a) installation; (b) 
remeasurement (reproduced with permission from Dunne, 1977). 

of the pin method, Sims et al. (1999) evaluated the effects of barriers on 
soil retention with the help of permanent benchmarks above and below 
them. Measurements were made to the soil surface from a taut cable tied 
to the benchmarks, which indicated the accumulation or loss of soil before 
and after each rainy season. This was combined with annual measurements 
of slope of the interbarrier soil to indicate changes in microtopography. 

1 7.3 Qualitative Methods 

Tree root exposure or soil pedestals 

It is sometimes possible to reconstruct the recent erosion history of an area 
from truncations of soil profiles, from the height of residual soil pedestals, 
or from the exposure of tree roots. Normal spatial variations of soil profile 
depth with slope gradient must be taken into consideration. If soil erosion 
is rapid following the destruction of vegetative cover, remnants of the 
former surface may be left. A ruler, tape or frame laid across the former 
surface can be used as a reference from which to measure the average 
depth of erosion (Fig. 17.4a). The exposure of tree roots can also be 
measured (Fig. 17.4b). However, some trees, even on undisturbed sites, 
grow with part of their roots above the ground surface. So, before such 
measurements are made, trees of the same species in neighbouring areas 
should be investigated. 

Changes in soil microtopography 

Evidence of the breakdown of the surface soil structure and its subsequent 
transport across the soil surface can be seen in the change in microtopo- 


M.A. McDonald et al. 



of erosion . 


t Depth of erosion 

/pen Level 

Minimum level of former 

soil surface •. 

Present soil 

Fig. 17.4. (a) Measurement of erosion between vegetated remnants of former soil 
surface; (b) measurement of erosion around tree roots (reproduced with permission 
from Dunne, 1 977). 

graphical features of the soil surface. These include the breakdown of clods 
formed by ploughing; the filling of ploughed depressions by soil loosened 
from the clods and from the rest of the soil surface; and indicators that 
water has flowed across the soil, such as flowlines and rills. Ranking 
schemes can be developed to undertake series of structured field 
observations of the evidence for soil loss provided by these features 
(Bergsma, 1992). Stocking and Murnaghan (2001) suggest developing a 
context-specific ranking scheme which, if used consistently, can be a good 
way of combining indicators to give a more comprehensive view of land 
degradation. For example, to assess rill erosion, a scale of 0-3 indicates the 
development from no rills present to the presence of deep rills (up to 300 
mm depth) and/or rills affecting more than 25% of the surface area. The 
FAO (1976) advocates wider ranking for the same parameter with a scale 
from to 14, which describes a range from no visual evidence of rills, to 
rills present 7.5-15 cm deep at intervals less than 1.5 m. Utilization of 
numerically wider ranges has the added advantage of being more sensitive 
to the determination of statistical significance. The FAO (1976) and 
Bergsma (1992) apply such rankings to a range of topographical features 
including flow patterns, gullies, surface litter accumulation and 
depressions. Sources of error can include classification of the features, a 
low frequency of observations, and mislocating the same location for 
repeated recordings (Bergsma, 1992). Special attention should be given 

Soil Erosion 343 

to signs of runoff and transport of eroded soil at the field boundaries, 
where soil is effectively lost from the field, i.e. on the down-slope side in 
all fields and on the lateral sides in contour-ridged and terraced fields. 

Stable isotopes 

Caesium- 137 is a by-product of the nuclear bomb testing that occurred in 
the 1950s and 1960s. It has a relatively long half-life of 30.17 years and 
does not occur naturally in the environment, so can be used as an indicator 
of disturbance if comparison is made between undisturbed areas and those 
subject to perturbation by erosion. The technique has been applied 
successfully to terraced areas in China and Nepal (Quine et al., 1992; Zhang 
et al, 1994; Gardner et al., 2000). Gardner et al. (2000) noted very high 
spatial variability, with the worst cases indicating losses roughly equivalent 
to 40-50% of the top 10 cm of soil over the 40-year period. They concluded 
that there was no systematic spatial pattern of net loss or accumulation 
down-slope associated with natural convexities or concavities in the slope 
topography, and that the differences were more likely to reflect localized 
pathways of preferential drainage and water movement. Their results 
highlight the processes of redistribution of soil on a hillslope scale, and 
indicate the usefulness of this technique as a relative tool for qualitative 
assessment of soil erosion to indicate patterns of spatial loss and 
accumulation. The extension of the technique to provide quantitative 
estimates of loss remains to be developed (Walling and Quine, 1990). 
Estimates of soil loss using caesium-137 should be compared with other 
means since they may be subject to errors arising from uneven mixing of 
the cultivated layer and preferential adsorption of the isotope on to clay 
particles during the processes of erosion, transport and deposition 
(Morgan, 1995). 


Abbas, J. D., Hetrick, B.A.D. and Jurgenson, J.E. (1996) Isolate specific detection 

of mycorrhizal fungi using genome specific primer pairs. Mycologia 88, 

Abe, T. (1987) Evolution of life types in termites. In: Kawano, S., Connell, J.H. 

and Hidaka, T. (eds) Evolution and Coadaptation in Biotic Communities. University 

of Tokyo Press, Tokyo, pp. 125-148. 
Acquaye, D.F. and Kang, B.T. (1987) Sulfur status and forms in some surface soils 

of Ghana. Soil Science 144, 43-52. 
Adderley, W.P., Jenkins, D.A., Sinclair, F.L., Stevens, PA. and Verinumbe, I. 

(1997) The influence of soil variability on tree establishment at an 

experimental agroforestry site in North East Nigeria. Soil Use and Management 

13, 1-8. 
Adesina, A.A. and Coulibaly, O.N. (1998) Policy and competitiveness of 

agroforestry-based technologies for maize production in Cameroon: an 

application of policy analysis matrix. Agricultural Economics 19, 1-13. 
Adesina, A.A., Mbila, D., Nkamleu, G.B. and Endamana, D. (2000) Econometric 

analysis of the determinants of adoption of alley farming by farmers in the 

forest zone of southwest Cameroon. Agriculture, Ecosystems and Environment 80, 

Agbenin, J.O., Iwuafor, E.N.O. and Ayuba, B. (1999) A critical assessment of 

methods for determining organic phosphorus in savanna soil. Biology and 

Fertility of Soils 28, 177-181. 
Agbim, N.N. (1987) Dry season decomposition of leaf litter from five common plant 

species of West Africa. Biological Agriculture and Horticulture 4,21 3-224. 
Agerer, R. (1992) Characterization of ectomycorrhiza. Methods in Microbiology 23, 

Agerer, R. (1993) Colour Atlas of Ectomycorrhizas. Einhorn, Schwabisch Gmiind. 


346 Refer. 

Agus, F., Cassel, D.K. and Garrity, D.P. (1997) Soil water and soil physical 

properties under contour hedgerow systems on sloping Oxisols. Soil and Tillage 

Research 40, 185-199. 
Agustin, E.O. and Nortcliff, S. (1994) Agroforestry practices to control runoff and 

erosion in Ilocos Norte, Philippines. In: Syres, J.K. and Rimmer, D.L. (eds) 

Soil Science and Sustainable Land Management in the Tropics. CAB International, 

Wallingford, UK, pp. 59-72. 
Ahenkorah, Y. (1975) Use of radio-active phosphorus in determining the efficiency 

of fertilizer utilization by cacao plantation. Plant and Soil 42, 429-439. 
Ahenkorah, Y., Halm, B.J., Appiah, M.R., Akron, G.S. and Yirenkyi, J.E.K. (1987) 

Twenty years' results from a shade and fertilizer trial on Amazon cocoa 

(Theobroma cacao) in Ghana. Experimental Agriculture 23, 31-39. 
Ahuja, L.R., El-Swaify, SA. and Rahman, A. (1976) Measuring hydrologic 

properties of soil with a double-ring infiltrometer and multiple-depth 

tensiometers. Soil Science Society of America Journal 40, 494-499. 
Ahuja, L.R., Green, R.E., Chong, S.K. and Nielsen, D.R. (1980) A simplified 

functions approach for determining soil hydraulic conductivities and water 

characteristics in situ. Water Resources Research 16, 947-953. 
Ahuja, L.R., Barnes, B.B., Cassel, D.K., Bruce, R.R. and Nofziger, D.L. (1988) 

Effect of assumed unit gradient during drainage on the determination of 

unsaturated hydraulic conductivity and infiltration parameters. Soil Science 

Akin, H . and Siemes, H . ( 1 988) Praktische Geostatistik, eine Einfuhrung fur den Bergbau 

und die Geowissenschaften. Springer, Berlin, 304 pp. 
Akinnifesi, F.K., Kang, B.T. and Lapido, D.O. (1999) Structural root form and 

fine root distribution of some woody species evaluated for agroforestry 

systems. Agroforestry Systems 42, 121-138. 
Alef, K. and Nannipieri, P. (1995) Methods in Applied Soil Microbiology and 

Biochemistry. Academic Press, London, 512 pp. 
Alegre, J.C. and Rao, M.R. (1996) Soil and water conservation by contour hedging 

in the humid tropics of Peru. Agriculture, Ecosystems and Environment 57, 17-25. 
Alegre, J.C, Cassel, D.K. and Bandy, D.E. (1986) Reclamation of an Ultisol 

damaged by mechanical land clearing. Soil Science Society of America Journal 50, 

Alfaro, M. (1980) The random coin method: solution of the problem of the 

simulation of a random function in the plane. Mathematical Geology 12, 25-32. 
Allen, M.F. (1991) The Ecology of Mycorrhizae. Cambridge University Press, 

Cambridge, 184 pp. 
Allen, M.F., Moore, T.S. and Christensen, M. (1982) Phytohormone changes in 

Bouteloua gracilis infected by vesicular-arbuscular mycorrhizae. 2. Altered levels 

of gibberellin-like substances and abscisic acid in the host plant. Canadian 

Journal of Botany 60, 468-471. 
Allen, O.N. and Allen, E.K. (1981) The Leguminosae: a Source Book of Characteristics, 

Uses and Nodulation. University of Wisconsin, Madison, Wisconsin, 812 pp. 
Allen, T.F.H. and Starr, T.B. (1982) Hierarchy Perspectives for Ecological Complexity. 

University of Chicago Press, Chicago, Illinois, 310 pp. 
Alphei, J., Bonkowski, M. and Scheu, S. (1996) Protozoa, Nematoda and 

Lumbricidae in the rhizosphere of Hordelymus europaeus (Poaceae): faunal 

References 347 

interactions, response of microorganisms and effects on plant growth. Oecologia 

106, 111-126. 
Amelung, W., Zech, W., Zhang, X., Follett, R.F., Tiessen, H., Knox, E. and Flach, 

K.W. (1998) Carbon, nitrogen and sulfur pools in particle size fractions as 

influenced by climate. Soil Science Society of America Journal 62, 172-181. 
Anderson, D.W. and Schoenau, J.J. (1993) Soil humus fractions. In: Carter, M.R. 

(ed.) Soil Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, 

Florida, pp. 391-395. 
Anderson, D.W., Saggar, S., Bettany, J.R. and Stewart, J. W.B. (1981) Particle size 

fractions and their use in studies of soil organic matter: I. The nature and 

distribution of forms of carbon, nitrogen and sulfur. Soil Science Society of 

America Journal 45, 767-772. 
Anderson, J.M. and Ingram, J.S.I. (1993) Tropical Soil Biology and Fertility: a 

Handbook of Methods. CAB International, Wallingford, UK, 221 pp. 
Andren, O. and Katterer, T. (1997) ICBM - the Introductory Carbon Balance 

Model for exploration of soil carbon balances. Ecological Applications 7, 

Angers, D.A. and Caron, J. (1998) Plant-induced changes in soil structure: 

processes and feedbacks. Bio geochemistry 42, 55-72. 
Angers, D.A. and Mehuys, G.R. (1993) Aggregate stability to water. In: Carter, 

M.R. (ed.) Soil Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, 

Florida, pp. 651-657. 
Angle, J. S., Mcintosh, M.S. and Hill, R.L. (1991) Tension lysimeters for collecting 

soil percolate. In: Nash, R.G. and Leslie, A.R. (eds) Groundwater Residue 

Sampling Designs. American Chemical Society, Washington, DC, pp. 290-299. 
Ankeny, M.D. (1992) Methods and theory for unconfined infiltration 

measurements. In: Topp, G.C., Reynolds, W.D. and Green, R.E. (eds) Advances 

in Measurement of Soil Physical Properties: Bringing Theory into Practice. Soil 

Science Society of America, Madison, Wisconsin, pp. 123-141. 
Ankeny, M.D., Kaspar, T.C. and Horton, R. (1990) Characterization of tillage and 

traffic effects on unconfined infiltration measurements. Soil Science Society of 

America Journal 54, 837-840. 
Ankeny, M.D., Ahmed, M., Kaspar, T.C. and Horton, R. (1991) Simple field 

method for determining unsaturated hydraulic conductivity. Soil Science Society 

of America Journal 55, 467-470. 
Archer, J.R. and Smith, P.D. (1972) The relation between bulk density, available 

water capacity, and air capacity of soils. Journal of Soil Science 23, 475-480. 
Armstrong, C.L. and Mitchell, J.K. (1988) Plant canopy characteristics and 

processes which affect transformation of rainfall properties. Transactions of the 

American Society of Agricultural Engineers 31, 1400-1409. 
Armstrong, W. (1994) Polarographic oxygen electrodes and their use in plant 

aeration studies. Proceedings of the Royal Society of Edinburgh Section B - Biological 

Sciences 102, 511-527. 
Arnold, J. E.M. (1997) Retrospect and prospect. In: Michael, J. E. and Dewees, P.A. 

(eds) Farms, Trees and Farmers. Earthscan Publications, London, pp. 271-287. 
Arya, L.M., Leij, F.J., van Genuchten, M.T. and Shouse, P.J. (1999) Scaling 

parameter to predict the soil water characteristics from particle-size 

distribution data. Soil Science Society of America Journal 63, 510-519. 

348 Refer. 

Ashwath, N., Dart, P.J., Edwards, D.G. and Khanna, P.K. (1995) Tolerance of 

Australian tropical and subtropical Acacias to acid soil. Plant and Soil 171, 

Asseline, J. and Valentin, C. (1978) Construction et mise au point d'un 

infiltrometre a aspersion. Cahiers ORSTOM, serie Pedologie 15, 321-348. 
Atayese, M.O., Awotoye, O.O., Osonubi, O. and Mulongoy, K. (1993) Comparisons 

of the influence of vesicular-arbuscular mycorrhiza on the productivity of 

hedgerow woody legumes and cassava at the top and the base of a hillslope in 

alley cropping systems. Biology and Fertility of Soils 16, 198-204. 
Atkinson, D. (1974) Some observations on the distribution of root activity in apple 

trees. Plant and Soil 40, 333-342. 
Atkinson, D., Johnson, M.G., Mattam, D. and Mercer, E.R. (1978) The effect of 

orchard soil management on the uptake of nitrogen by established apple trees. 

Journal of the Science of Food and Agriculture 30, 129-135. 
Auge, R.M., Schekel, K.A. and Wample, R.L. (1986) Greater leaf conductance of 

well-watered VA mycorrhizal role plants is not related to phosphorus 

nutrition. New Phytologist 103, 107-116. 
Ba, A.M., Plenchette, C, Danthu, P., Duponnois, R. and Guissou, T. (2000) 

Functional compatibility of two arbuscular mycorrhizae with thirteen fruit 

trees in Senegal. Agroforestry Systems 50, 95-105. 
Babbar, L.I. and Ewel, J.J. (1989) Descomposicion des follaje en diversos 

ecosistemas sucesionales tropicales. Biotropica 21, 20-29. 
Babu, S.C., Hallam, A. and Rajasekaran, B. (1995) Dynamic modelling of 

agroforestry and soil fertility interactions: implications for multi-disciplinary 

research policy. Agricultural Economics 13, 125-135. 
Bache, B.W. and Williams, E.G. (1971) A phosphate sorption indexfor soils. Journal 

of Soil Science 22, 289-301. 
Bagyaraj, D.J. (1994) Vesicular-arbuscular mycorrhiza: applications in agriculture. 

In: Norris, J.R., Read, D. and Varma, A.K. (eds) Techniques for Mycorrhizal 

Research. Academic Press, London, pp. 819-833. 
Bal, L. (1982) Zoological Ripening of Soils -the Concept, and Impact in Pedology, Forestry 

and Agriculture. Pudoc, Wageningen, 365 pp. 
Bala, A. (1999) Biodiversity of Rhizobia which nodulate fast-growing tree legumes 

in tropical soils. PhD thesis. Wye College, University of London, London, 

216 pp. 
Bala, A. and Giller, K.E. (2001) Symbiotic specificity of tropical tree rhizobia for 

host legumes. New Phytologist 149, 495-508. 
Balesdent, J. (1996) The significance of organic separates to carbon dynamics and 

its modelling in some cultivated soils. European Journal of Soil Science 47, 

Balesdent, J. and Balabane, M. (1996) Major contribution of roots to soil carbon 

storage inferred from maize cultivated soils. Soil Biology and Biochemistry 28, 

Ball-Coelho, B., Sampaio, E.V.S.B., Tiessen, H. and Stewart, J. W.B. (1992) Root 

dynamics in plant and ratoon crops of sugar cane. Plant and Soil 142, 297-305. 
Ball-Coelho, B., Salcedo, I.H., Tiessen, H. and Stewart, J.W.B. (1993) Short- and 

long-term phosphorus dynamics in a fertilized ultisol under sugarcane. Soil 

Science Society of America Journal 57, 1027-1034. 

References 349 

Barbee, G.C. and Brown, K.W. (1986) Comparison between suction and free- 
drainage soil solution samplers. Soil Science 141, 149-154. 
Barber, R.G. and Navarro, F. (1994) The rehabilitation of degraded soils in eastern 

Bolivia by subsoiling and the incorporation of cover crops. Land Degradation 

and Rehabilitation 5, 247-259. 
Barber, S.A. (1995) Soil Nutrient Bioavailability. John Wiley & Sons, London, 414 

Barbier, E.B. (1990) The farm-level economics of soil conservation: the uplands 

of Java. Land Economics 66, 199-211. 
Bardossy, A. (ed.) (1992) Geostatistical Methods: Recent Developments and Applications 

in Surface and Subsurface Hydrology . Unesco, Paris, 161 pp. 
Barea, J.M. and Azcon-Aguilar, C. (1982) Production of plant growth regulating 

substances by the vesicular-arbuscular mycorrhizal fungus Glomus mosseae. 

Applied and Environmental Microbiology 43, 810-813. 
Barea, J.M. and Jeffries, P. (1995) Arbuscular mycorrhizas in sustainable soil-plant 

systems. In: Hock, B. and Varma, A. (eds) Mycorrhiza - Structure, Function, 

Molecular Biology and Biotechnology. Springer, Berlin, pp. 521-560. 
Barraclough, D. (1995) 15 N isotope dilution techniques to study soil nitrogen 

transformations and plant uptake. Fertilizer Research 42, 185-192. 
Barrie, A., Brookes, S.T., Prosser, S.J. and Debney, S. (1995) High production 

analysis of 15 N and 13 C in soil/plant research. Fertilizer Research 42, 43-59. 
Barrios, E. and Herrera, R. (1994) Nitrogen cycling in a Venezuelan tropical 

seasonally flooded forest: soil nitrogen mineralization and nitrification. Journal 

of Tropical Ecology 10, 399-416. 
Barrios, E., Buresh, R.J. and Sprent, J.I. (1996) Nitrogen mineralization in density 

fractions of soil organic matter from maize and legume cropping systems. Soil 

Biology and Biochemistry 28, 1459-1465. 
Barrios, E., Kwesiga, F., Buresh, R.J. and Sprent, J.I. (1997) Light fraction soil 

organic matter and available nitrogen following trees and maize. Soil Science 

Society of America Journal 61, 826-831. 
Barrios, E., Kwesiga, F., Buresh, R.J., Sprent, J.I. and Coe, R. (1998) Relating 

preseason soil nitrogen to maize yield in tree legume-maize rotations. Soil 

Science Society of America Journal 62, 1604-1609. 
Barros, E., Neves, A., Fernandes, E.C.M., Wandelli, E. and Lavelle, P. (1999) Soil 

macrofauna community of Amazonian agroforestry systems. In: International 

Symposium on Multi-strata Agroforestry Systems with Perennial Crops, 22-27 Feb. 

1999, Extended Abstracts. CATIE, Turrialba, pp. 166-170. 
Barros, E., Curmi, P., Hallaire, V., Chauvel, A. and Lavelle, P. (2001) The role of 

macrofauna in the transformation and reversibility of soil structure of an oxisol 

in the process of forest to pasture conversion. Geoderma 100, 193-213. 
Barton, L., McLay, C.D.A., Schipper, L.A. and Smith, C.T. (1999) Annual 

denitrification rates in agricultural and forest soils: a review. Australian Journal 

of Soil Research 37, 1073-1093. 
Bassett, D.M., Stockton, J. R. and Dickens, W.L. (1970) Root growth of cotton as 

measures by 32 P uptake. Agronomy Journal 62, 200-203. 
Batchelor, W.D. (1998) Fundamentals of neuronal networks. In: Peart, R.M. and 

Curry, R.B. (eds) Agricultural Systems - Modeling and Simulation. Marcel Dekker, 

New York, pp. 597-628. 

350 Refer. 

Bauhus, J. (1998) Does the abscission of fine roots lead to immobilization of 

nitrogen in microbial biomass during in situ soil nitrogen mineralization 

measurements? Communications in Soil Science and Plant Analysis 29, 1007-1022. 
Bazzanella, A., Lochmann, H., Tomos, A.D. and Bachmann, K. (1998) 

Determination of inorganic cations and anions in single plant cells by capillary 

zone electrophoresis. Journal of Chromatography, A 809, 231-239. 
Beck, M.A. and Sanchez, P.A. (1996) Soil phosphorus movement and budget 

after 13 years of fertilized cultivation in the Amazon basin. Plant and Soil 184, 

Beck, T., Joergensen, R.G., Kandeler, E., Makeschin, F., Nuss, E., Oberholzer, 

H.R. and Scheu, S. (1997) An inter-laboratory comparison often different 

ways of measuring soil microbial biomass C. Soil Biology and Biochemistry 29, 

Becking, J. H. (1992) The Rhizobium symbiosis of the nonlegume Parasponia. In: 

Stacey, G., Burris, R.H. and Evans, H.J. (eds) Biological Nitrogen Fixation. 

Chapman and Hall, New York, pp. 497-559. 
Beer, J. (1988) Litter production and nutrient cycling in coffee (Coffea arabica) or 

cacao (Theobroma cacao) plantations with shade trees. Agroforestry Systems 7, 

Beer, J., Muschler, R., Kass, D. and Somarriba, E. (1998) Shade management in 

coffee and cacao plantations. Agroforestry Systems 38, 139-164. 
Beier, C. and Hansen, K. (1992) Evaluation of porous cup soil-water samplers 

under controlled field conditions: comparison of ceramic and PTFE cups. 

Journal of Soil Science 43, 261-271. 
Bell, M.J., Moody, P.W., Connolly, R.D. and Bridge, B.J. (1998) The role of active 

fractions of soil organic matter in physical and chemical fertility of Ferrosols. 

Australian Journal of Soil Research 36, 809-819. 
Bell, M.J., Moody, P.W., Yo, SA. and Connolly, R.D. (1999) Using active fractions 

of soil organic matter as indicators of the sustainability of Ferrosol farming 

systems. Australian Journal of Soil Research 37, 279-287. 
Benjaminsen, T.A. (1998) Beyond 'Degradation': Essays on People, Lund and Resources 

in Mali. Centre for Development and the Environment, University of Oslo, 

Oslo, 140 pp. 
Bereau, M., Gazel, M. and Garbaye, J. (1997) Les symbioses mycorrhiziennes des 

arbres de la foret tropical humide de Guyane franchise. Canadian Journal of 

Botany 75, 711-716. 
Bergmann, W. (1988) Erndhrungsstorungen bei Kulturpflanzen. Gustav Fischer, 

Stuttgart, 762 pp. 
Bergsma, E. (1992) Features of soil surface microtopography for erosion hazard 

evaluation. In: Hurni, H. and Tato, K. (eds) Erosion, Conservation and Small 

Scale Farming. Geographica Bernensia, Bern, pp. 15-26. 
Bernhard-Reversat, F. (1993) Dynamics of litter and organic matter at the soil-litter 

interface in fast-growing tree plantations on sandy ferrallitic soils (Congo). 

Acta Oecologica 14, 179-195. 
Bernhard-Reversat, F. (1998) Changes in relationships between initial litter quality 

and CO, release during early laboratory decomposition of tropical leaf litters. 

European Journal of Soil Biology 34, 1 17-122. 
Bernoux, M., Cerri, C.C., Neill, C. and de Moraes, J.F.L. (1998) The use of stable 

References 351 

carbon isotopes for estimating soil organic matter turnover rates. Geoderma 82, 

Bethlenfalvay, G.J., Pacovsky, R.S. and Brown, M.S. (1981) Measurement of 

mycorrhizal infection in soybeans. Soil Science Society of America Journal 45, 

Bethlenfalvay, G.J., Brown, M.S., Ames, R.N. and Thomas, R.S. (1988) Effects 

of drought on host and endophyte development in mycorrhizal soybeans 

in relation to water use and phosphate uptake. Physiologia Plantarum 72, 

Bettany, J.R., Saggar, S. and Stewart, J. W.B. (1980) Comparison of the amount 

and forms of sulfur in soil organic matter fractions after 65 years of cultivation. 

Soil Science Society of America Journal 44, 70-74. 
Beven, K. (1991) Modeling preferential flow: an uncertain future? In: Gish, T.J. 

and Shirmohammadi, A. (eds) Preferential Flow. Proceedings of the National 

Symposium, 16-17 December 1991, Chicago, Illinois. American Society of 

Agricultural Engineers, Stjoseph, Minnesota, pp. 1—11. 
Binkley, D. (1984) Ion exchange resin bags: factors affecting estimates of nitrogen 

availability. Soil Science Society of America Journal 48, 1 181-1 184. 
Birch, H.F. (1960) Nitrification in soils after different periods of dryness. Plant and 

Soil 1,81-96. 
Bishop, K. and Dambrine, E. (1995) Localization of tree water uptake in Scots pine 

and Norway spruce with hydrological tracers. Canadian Journal of Forest 

Research 25, 286-297. 
Le Bissonnais, Y. (1996) Aggregate stability and assessment of soil crustability and 

erodibility: I. Theory and methodology. European Journal of Soil Science 47, 

Le Bissonnais, Y. and Singer, M.J. (1993) Seal formation, runoff and interrill 

erosion from seventeen California soils. Soil Science Society of America Journal 

Black, T.A., Gardner, W.R. and Thurtell, G.W. (1969) The prediction of 

evaporation, drainage and soil water storage for a bare soil. Soil Science Society 

of America Proceedings 33, 655-660. 
Blaikie, P. (1986) The Political Economy of Soil Erosion in Developing Countries. 

Longman, Harlow, 188 pp. 
Blaikie, P. and Brookfield, H. (1987) Land Degradation and Society. Methuen, 

London, 296 pp. 
Blair, G.J., Lefroy, R.D.B. and Lisle, L. (1995) Soil carbon fractions based on their 

degree of oxidation and the development of a carbon management index for 

agricultural systems. Australian Journal of Agricultural Research 46, 1459-1466. 
Blair, G.J., Lefroy, R.D.B. , Singh, B.P. and Till, A.R. (1997) Development and use 

of a carbon management index to monitor changes in soil C pool size and 

turnover rate. In: Cadisch, G. and Giller, K.E. (eds) Driven by Nature: Plant 

Litter Quality and Decomposition. CAB International, Wallingford, UK, pp. 

Blair, J. M., Parmelee, R.W. and Beare, M.H. (1990) Decay rates, nitrogen fluxes, 

and decomposer communities of single- and mixed-species foliar litter. Ecology 

71, 1976-1985. 
Blair, J.M., Crossley, DA. Jr and Callaham, L.C. (1991) A litterbasket technique 

352 Refer. 

for measurement of nutrient dynamics in forest floors. Agriculture, Ecosystems 

and Environment 34, 465-471. 
Blake, G.R. and Hartge, K.H. (1986) Bulk density. In: Klute, A. (ed.) Methods of 

Soil Analysis, Part 1. American Society of Agronomy, Madison, Wisconsin, pp. 

Blanchart, E. (1992) Restoration by earthworms (Megascolecidae) of the 

macroaggregate structure of a destructured savanna soil under field 

conditions. Soil Biology and Biochemistry 24, 1587-1594. 
Blanchart, E., Bruand, A. and Lavelle, P. (1993) The physical structure of casts of 

Millsonia anomala (Oligochaeta: Megascolecidae) in shrub savanna soils (Cote 

d'lvoire). Geoderma 56, 119-132. 
Blanchart, E., Albrecht, A., Alegre, J.C., Duboisset, A., Pashanasi, B., Lavelle, P. 

and Brussaard, L. (1999) Effects of earthworms on soil structure and physical 

properties. In: Lavelle, P., Brussaard, L. and Hendrix, P. (eds) Earthworm 

Management in Tropical Agroecosystems. CAB International, Wallingford, UK, 

pp. 139-162. 
Blanchart, E., Albrecht, A., Decaens, T., Duboisset, A., Lavelle, P., Mariani, L. and 

Roose, E. (2003) Effects of tropical endogeic earthworms on soil erosion: a 

review. Agriculture, Ecosystems and Environment (in press). 
Blum, U., Shafer, S.R. and Lehman, M.E. (1999) Evidence for inhibitory 

allelopathic interactions involving phenolic acids in field soils: concepts vs. an 

experimental model. Critical Reviews in Plant Science 18, 673-693. 
Boddey, R.M. (1995) Biological nitrogen fixation in sugar cane: a key to 

energetically viable biofuel production. Critical Reviews in Plant Science 14, 

Boddey, R.M., Urquiaga, S. and Neves, M.C.P. (1990) Quantification of the 

contribution of N 9 -fixation to field grown legumes - a strategy for the practical 

application of the 15 N isotope dilution technique. Soil Biology and Biochemistry 

22, 649-655. 
Boddey, R.M., Peoples, M.B., Palmer, B. and Dart, P.J. (2000) Use of the 15 N 

natural abundance method to quantify biological nitrogen fixation by woody 

perennials. Nutrient Cycling in Agroecosystems 57, 235-270. 
Boettcher, S.E. and Kalisz, P.J. (1991) Single-tree influence on earthworms in forest 

soils in Eastern Kentucky. Soil Science Society of America Journal 55, 862-865. 
Boffa, J.M. (1999) Agroforestry Parklands in Sub-Saharan Africa. FAO Conservation 

Guide 34. Food and Agriculture Organization of the United Nations, Rome, 

230 pp. 
Bohm, W. (1979) Methods of Studying Root Systems. Springer, Berlin. 
Bojappa, K.M. and Singh, R.N. (1974) Root activity of mango by radiotracer 

technique using 32 P. Indian Journal of Agricultural Sciences 44, 175-180. 
Borowicz, V.A. (2001) Do arbuscular mycorrhizal fungi alter plant-pathogen 

relations? Ecology 82, 3057-3068. 
Bouche, M.B. (1977) Strategies lombriciennes. Ecological Bulletin (Stockholm) 25, 

Boutton, T.W. (1991) Stable carbon isotope ratios of natural materials I. Sample 

preparation and mass-spectrometric analysis. In: Coleman, D.C. and Frey, B. 

(eds) Carbon Isotope Techniques. Academic Press, New York, pp. 155-171. 
Boutton, T.W. (1996) Stable carbon isotope ratios of soil organic matter and their 

References 353 

use as indicators of vegetation and climate change. In: Boutton, T.W. and 

Yamasaki, S. (eds) Mass Spectrometry of Soils. Marcel Dekker, New York, pp. 

Boutton, T.W., Archer, S.R., Midwood, A.J., Zitzer, S.F. and Bol, R. (1998) 5 13 C 

values of soil organic carbon and their use in documenting vegetation change 

in a subtropical savanna ecosystem. Geoderma 82, 5-41. 
Bouwer, H. (1986) Intake rate: cylinder infiltrometer. In: Klute, A. (ed.) Methods 

of Soil Analysis, Part 1. American Society of Agronomy, Madison, Wisconsin, 

pp. 825-844. 
Bowen, G.D. (1984) Tree roots and the use of soil nutrients. In: Bowen, G.D. and 

Nambiar, E.K.S. (eds) Nutrition of Plantation Forests. Academic Press, London, 

pp. 147-179. 
Bowman, R.A., Rodriguez, J. B. and Self, J. R. (1998) Comparison of methods to 

estimate occluded and resistant soil phosphorus. Soil Science Society of America 

Journal 62, 338-342. 
Boyer, J. (1998) Interactions biologiques (faune, ravageurs, parasites, microflore) 

dans des sols sous cultures en milieu tropical humide (He de la Reunion). PhD 

thesis, Universite Paris VI, Paris, 246 pp. 
Bragato, G. and Primavera, F. (1998) Manuring and soil type influence on spatial 

variation of soil organic matter properties. Soil Science Society of America Journal 

Brandes, B., Godbold, D.L. and Jentschke, G. (1998) Nitrogen and phosphorus 

acquisition by the mycelium of the ectomycorrhizal fungus Paxillus involutus 

and its effect on host nutrition. New Phytologist 140, 735-744. 
Braum, S.M. and Helmke, PA. (1995) White lupin utilizes soil phosphorus that is 

unavailable to soybean. Plant and Soil 176, 95-100. 
Breman, H. and Kessler, J.J. (1997) The potential benefits of agroforestry in the 

Sahel and other semi-arid regions. European Journal of Agronomy 7, 25-33. 
Bremner, J.M. and Edwards, A.P. (1965) Determination and isotope-ratio analysis 

of different forms of nitrogen in soils: I. Apparatus and procedure for 

distillation and determination of ammonium. Soil Science Society of America 

Proceedings 29, 504-507. 
Brenner, A.J., Jarvis, P.G. and Vandenbeldt, R.J. (1995) Windbreak-crop 

interactions in the Sahel 2. Growth-response of millet in shelter. Agricultural 

and Forest Meteorology 75, 235-262. 
Brewer, R. (1964) Fabric and Mineral Analysis of Soils. John Wiley & Sons, New York, 

470 pp. 
Briones, M.J.I, and Ineson, P. (1996) Decomposition of eucalyptus leaves in litter 

mixtures. Soil Biology and Biochemistry 28, 1381-1388. 
Broeshart, A. and Nethsinghe, DA. (1972) Studies on the pattern of root activity 

of tree crops using isotope techniques. In: Isotopes and Radiation in Soil Plant 

Relationships Including Forestry. International Atomic Energy Agency, Vienna, 

pp. 453-463. 
Brookes, P.C., Powlson, D.S. and Jenkinson, D.S. (1982) Measurement of microbial 

biomass phosphorus in soil. Soil Biology and Biochemistry 14, 319-329. 
Brookes, P.C., Powlson, D.S. and Jenkinson, D.S. (1984) Phosphorus in the soil 

microbial biomass. Soil Biology and Biochemistry 16, 169-175. 
Brooks, P.D., Stark, J.M. , Mclnteer, B.B. and Preston, T. (1989) Diffusion method 

354 Refer. 

to prepare soil extracts for automated nitrogen- 15 analysis. Soil Science Society 

of America Journal 53 , 1707-1711. 
Broughton, W.J. (1977) Effect of various covers on soil fertility under 

Hevea brasiliensis Muell. Arg. and on growth of the tree. Agro-Ecosystems 3, 

Brown, J.K. and Roussopoulos, P.J. (1974) Eliminating biases in the planar 

intersect method for estimating volumes of small fuels. Forest Science 20, 

Bruand, A. (1990) Improved prediction of water retention properties of clayey 

soils by pedological stratification. Journal of Soil Science 38, 461-472. 
Bruand, A., Cousin, I., Nicoullaud, B., Duval, O. and Begon, J.C. (1996) 

Backscattered electron scanning images of soil porosity for analysing 

soil compaction around roots. Soil Science Society of America Journal 60, 

Bruijnzeel, LA. (1991) Nutrient input-output budgets of tropical forest 

ecosystems: a review . Journal of Tropical Ecology 7, 1-24. 
Bruijnzeel, LA. (1997) Hydrology of forest plantations in the tropics. In: Nambiar, 

E.K.S. and Brown, A.G. (eds) Management of Soil, Nutrients and Water in Tropical 

Forest Plantations. ACIAR, Canberra, pp. 125-167. 
Bruijnzeel, LA. and Critchley, W.R.S. (1996) A new approach towards the 

quantification of runoff and eroded sediment from bench terraces in humid 

tropical steeplands and its application in south-central Java, Indonesia. In: 

Anderson, M.G. and Brooks, S.M. (eds) Advances in Hillslope Processes, Vol. 2. 

Wiley & Sons, Chichester, UK, pp. 92 1-937. 
Brundrett, M.C. and Abbott, L.K. (1994) Mycorrhizal fungus propagules in the 

Jarrah forest. 1. Seasonal study of inoculum levels. New Phytologist 127, 

Brundrett, M.C, Piche, Y. and Peterson, R.L. (1984) A new method for observing 

the morphology of vesicular-arbuscular mycorrhizae. Canadian Journal of 

Botany 62,2128-2134. 
Brundrett, M., Bougher, N., Dell, B., Grove, T. and Malajczuk, N. (1996) Working 

with Mycorrhizas in Forestry and Agriculture . Australian Centre for International 

Agricultural Research, Canberra, 374 pp. 
Budelman, A. (1988) The decomposition of the leaf mulches of Leucaena 

leucocephala, Gliricidia sepium and Flemingia macrophylla under humid tropical 

conditions. Agroforestry Systems 7, 33-45. 
Budelman, A. (1989) Nutrient composition of the leaf biomass of three selected 

woody leguminous species. Agroforestry Systems 8, 39-51. 
Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G. and Tursina, T. (1985) Handbook 

of Soil Thin Section Description. Waine Research Publications, Wolverhampton, 

UK, 152 pp. 
Bunch, R. (1999) Reasons for non-adoption of soil conservation technologies and 

how to overcome them. Mountain Research and Development 19, 213-219. 
Bundt, M., Widmer, F., Pesaro, M., Zeyer, J. and Blaser, P. (2001) Preferential 

flow paths: biological 'hot spots' in soils. Soil Biology and Biochemistry 33, 

Buresh, R.J. (1999) Phosphorus management in tropical agroforestry: current 

knowledge and research challenges. Agroforestry Forum 9, 61-66. 

References 355 

Buresh, R.J. and Tian, G. (1998) Soil improvement by trees in sub-Saharan Africa. 

Agroforestry Systems 38, 51-76. 
Buresh, R.J., Austin, E.R. and Craswell, E.T. (1982) Analytical methods in 15 N 

research. Fertilizer Research 3, 37-62. 
Buresh, R.J., Smithson, P.C. and Heliums, D.T. (1997) Building soil phosphorus 

capital in Africa. In: Buresh, R.J., Sanchez, PA. and Calhoun, F. (eds) 

Replenishing Soil Fertility in Africa. Soil Science Society of America, Madison, 

Wisconsin, pp. 111-149. 
Burghouts, T.B.A., van Straalen, N.M. and Bruijnzeel, LA. (1998) Spatial 

heterogeneity of element and litter turnover in a Bornean rain forest. Journal 

of Tropical Ecology 14,477-506. 
Burke, M.K. and Raynal, D.J. (1994) Fine root growth phenology, production, 

and turnover in a northern hardwood forest ecosystem. Plant and Soil 162, 

Burkhardt, J. and Eiden, R. (1990) The ion concentration of dew condensed on 

Norway spruce (Picea abies (L.) Karst.) and Scots pine (Pinus sylvestris L.) 

needles. Trees 4, 22-26. 
Burkhardt, J. and Eiden, R. (1994) Thin water films on coniferous needles. 

Atmospheric Environment 28, 2001-2017. 
Busacca, A.J., Cook, CA. and Mulla, D.J. (1993) Comparing landscape-scale 

estimation of soil erosion in the Palouse using Cs-137 and RUSLE. Journal of 

Soil and Water Conservation 48, 361-367. 
Cadisch, G. and Ciller, K.E. (1996) Estimating the contribution of legumes to soil 

organic matter build up in mixed communities of C„/C 4 plants. Soil Biology and 

Biochemistry 28, 823-825. 
Cahn, M.D., Bouldin, D.R. and Cravo, M.S. (1992) Nitrate sorption in the profile 

of an acid soil. Plant and Soil 143, 179-183. 
Cahn, M.D., Bouldin, D.R., Cravo, M.S. and Bowen, W.T. (1993) Cation and nitrate 

leaching in an oxisol of the Brazilian Amazon. Agronomy Journal 85, 334-340. 
Cajas-Giron, Y.S. and Sinclair, F.L. (2001) Characterization of multistrata 

silvopastoral systems on seasonally dry pastures in the Caribbean region of 

Colombia. Agroforestry Systems 53, 215-225. 
Calba, H., Cazevieille, P. and Jaillard, B. (1999) Modelling of the dynamics of Al 

and protons in the rhizosphere of maize cultivated in acid substrate. Plant and 

Soil 209, 57-69. 
Calder, I.R. (2001) Canopy processes: implications for transpiration, interception 

and splash induced erosion, ultimately for forest management and water 

resources. Plant Ecology 153, 203-214. 
Caldwell, M.M. and Richards, J. H. (1989) Hydraulic lift: water efflux from upper 

roots improves effectiveness of water uptake by deep roots. Oecologia 79, 1-5. 
Caldwell, M.M. and Virginia, RA. (1989) Root systems. In: Pearcy, R.W., 

Ehleringer, J.R., Mooney, HA. and Rundel, P.W. (eds) Plant Physiological 

Ecology. Chapman and Hall, London, pp. 367-398. 
Caliman, J.P., Olivin, J. and Dufour, O. (1987) Degradation des sols ferralitiques 

sableux en culture de palmiers a huile par acidification et compaction. 

Oleagineux 42, 393-399. 
Callaway, R.M. (1995) Positive interactions among plants. Botanical Review 61, 


356 Refer. 

Callaway, R.M. and Walker, L.R. (1997) Competition and facilitation: a synthetic 

approach to interactions in plant communities. Ecology 78, 1958-1965. 
Cameron, K.C. and Haynes, R.J. (1986) Retention and movement of nitrogen in 

soils. In: Haynes, R.J. (ed.) Mineral Nitrogen in the Plant-Soil System. Academic 

Press, Orlando, Florida, pp. 166-241. 
Campbell, C.A., Ellert, B.H. and Jame, Y.W. (1993) Nitrogen mineralization 

potential in soils. In: Carter, M.R. (ed.) Soil Sampling and Methods of Analysis. 

Lewis Publishers, Boca Raton, Florida, pp. 341-349. 
Campbell, C.A.,Jame, Y.W., Akinremi, O.O. and Cabrera, M.L. (1995) Adapting 

the potentially mineralizable N concept for the prediction of fertilizer N 

requirements. Fertilizer Research 42, 61-75. 
Campbell, CD., Mackie-Dawson, LA., Reid, E.J., Pratt, S.M., Duff, E.I. and 

Buckland, S.T. (1994) Manual recording of minirhizotron data and its 

application to study the effect of herbicide and nitrogen fertiliser on tree and 

pasture root growth in a silvopastoral system. Agroforestry Systems 26, 75-87. 
Campbell, G.S. (1974) A simple method for determining unsaturated conductivity 

from moisture retention data. Soil Science 117, 311-314. 
Campbell, G.S. and Gee, G.W. (1986) Water potential: miscellaneous methods. In: 

Klute, A. (ed.) Methods of Soil Analysis, Part 1. American Society of Agronomy, 

Madison, Wisconsin, pp. 597-618. 
Campo, J., Jaramillo, V.J. and Maass, J.M. (1998) Pulses of soil phosphorus 

availability in a Mexican tropical dry forest: effects of seasonality and level of 

wetting. Oecologia 115, 167-172. 
Canadell, J., Jackson, R.B., Ehleringer, J.R., Mooney, H.A., Sala, O.E. and Schulze, 

E.-D. (1996) Maximum rooting depth of vegetation types at the global scale. 

Oecologia 108, 583-595. 
Cape, J.N. (1993) The use of 3o S to study sulphur cycling in forests. Environmental 

Geochemistry and Health 15, 113-118. 
Carter, M.R. (1993) Soil Sampling and Methods of Analysis. Lewis Publishers, Boca 

Raton, Florida, 823 pp. 
Carter, M.R., Angers, DA. and Kunelius, H.T. (1994) Soil structural form and 

stability, and organic matter under cool-season perennial grasses. Soil Science 

Society of America Journal 58, 1 194-1 199. 
Carter, M.R., Gregorich, E.G., Angers, DA., Beare, M.H., Sparling, G.P., Wardle, 

DA. and Voroney, R.P. (1999) Interpretation of microbial biomass 

measurements for soil quality assessment in humid temperate regions. 

Canadian Journal of Soil Science 79, 507-520. 
Carter, S.E. and Murwira, H.K. (1995) Spatial variability in soil fertility 

management and crop response in Mutoko Communal Area, Zimbabwe. Ambio 

24, 77-84. 
Carvalho, JA. Jr, Santos, J. M., Santos, J. C, Leitao, M.M. and Higuchi, N. (1995) 

A tropical rainforest clearing experiment by biomass burning in the Manaus 

region. Atmospheric Environment 29, 2301-2309. 
Cassel, D.K. and Klute, A. (1986) Water potential: tensiometry. In: Klute, A. (ed.) 

Methods of Soil Analysis, Part 1. American Society of Agronomy, Madison, 

Wisconsin, pp. 563-596. 
Cavelier, J. (1996) Environmental factors and ecophysiological processes along 

altitudinal gradients in wet tropical mountains. In: Mulkey, S.S., Chazdon, 

References 357 

R.L. and Smith, A.P. (eds) Tropical Forest Plant Ecophysiology. Chapman and 

Hall, New York, pp. 399-439. 
Ceccotti, S.P., Morris, R.J. and Messik, D.L. (1998) A global overview of the sulphur 

situation: industry's background, market trends, and commercial aspects of 

sulphur fertilizers. In: Schnug, E. (ed.) Sulphur in Agroecosystems. Kluwer 

Academic Publishers, Dordrecht, pp. 175-202. 
Chalk, P.M. and Ladha, J.K. (1999) Estimation of legume symbiotic dependence: 

an evaluation based on 15 N dilution. Soil Biology and Biochemistry 31, 1901-1917. 
Chambers, J. Q., Higuchi, N., Schimel, J.P., Ferreira, L.V. and Melack, J.M. (2000) 

Decomposition and carbon cycling of dead trees in tropical forests of the 

central Amazon. Oecologia 122, 380-388. 
Chambers, J. Q., Schimel, J. P. and Nobre, A.D. (2001) Respiration from coarse 

wood litter in central Amazon forests. Bio geochemistry 52, 115-131. 
Chapman, K., Whittaker, J.B. and Heal, O.W. (1988) Metabolic and faunal activity 

in litter of tree mixtures compared with pure stands. Agriculture, Ecosystems and 

Environment 24, 33-40. 
Chauvel, A., Lucas, Y. and Boulet, R. (1987) On the genesis of the soil mantle of 

the region of Manaus, Central Amazonia, Brazil. Experientia 43, 234-241. 
Chauvel, A., Grimaldi, M. and Tessier, D. (1991) Changes in soil pore-space 

distribution following deforestation and revegetation: an example from the 

Central Amazon Basin, Brazil. Forest Ecology and Management 38, 259-271. 
Chauvel, A., Grimaldi, M., Barros, E., Blanchart, E., Desjardins, T., Sarrazin, M. 

and Lavelle, P. (1999) An Amazonian earthworm compacts more than a 

bulldozer. Nature 398, 32-33. 
Chelius, M.K. and Lepo, J.E. (1999) Restriction fragment length polymorphism 

analysis of PCR-amplified nifH sequences from wetland plant rhizosphere 

communities. Environmental Technology 20, 883-889. 
Chellerni, D.O., Rohrbach, K.G., Yost, R.S. and Sonoda, R.M. (1988) Analysis of 

the spatial patterns of plant pathogens and diseased plants using geostatistics. 

Phytopathology 78, 221-226. 
Chenu, C. (1989) Influence of a fungal polysaccharide, scleroglucan, on clay 

microstructures. Soil Biology and Biochemistry 21, 299-305. 
Cheshire, M.V. and Mundie, CM. (1990) Organic matter contributed to soil by plant 

roots during the growth and decomposition of maize . Plant and Soil 121, 107-114. 
Chirwa, P.W., Nair, P.K.R. and Kamara, C.S. (1994a) Soil moisture changes and 

maize productivity under alley cropping with Leucaena and Flemingia 

hedgerows at Chalimbana near Lusaka, Zambia. Forest Ecology and Management 

Chirwa, P.W., Nair, P.K.R. and Knedi-Kizza, P. (1994b) Pattern of soil moisture 

depletion in alley cropping under semiarid conditions in Zambia. Agroforestry 

Systems 26, 89-99. 
Clapp, J. P., Fitter, A.H. and Young, J.P.W. (1999) Ribosomal small subunit 

variation within spores of an arbuscular mycorrhizal fungus, Scutellospora sp. 

Molecular Ecology 8,91 5-92 1 . 
Clarholm, M. (1985) Interactions of bacteria, protozoa and plants leading to 

mineralization of soil nitrogen. Soil Biology and Biochemistry 17, 181-187. 
Clark, R.B. and Zeto, S.K. (2000) Mineral acquisition by arbuscular mycorrhizal 

plants. Journal of Plant Nutrition 23, 867-902. 

358 Refer. 

Clemens, A., Schillinger, M.P., Goldbach, H. and Huwe, B. (1999) Spatial 

variability of N 2 emissions and soil parameters on an arable silty loam - a 

field study. Biology and Fertility of Soils 28, 403-406. 
Cochran, G. and Cox, M.C. (1957) Experimental Designs, 2nd edn. John Wiley & 

Sons, New York, 611 pp. 
Coe, R. (1997a) Checklist for planning on station agroforestry experiments. 

Available at: .htm 
Coe, R. (1997b) Plot structure. Available at: 

Coe, R. (1997c) Selecting sites. Available at: 

Coe, R. (1999) Experiments with Farmers. Checklist for Preparing Protocols. ICRAF, 

Nairobi, 5 pp. 
Colleuille, H. (1993) Approche physique et morphologique de la dynamique 

structurale des sols. Application a l'etude de deux sequences pedologiques 

tropicales. PhD thesis, University Paris VI. ORSTOM, Paris, 353 pp. 
Collis-George, N. (1977) Infiltration equations for simple soil systems. Water 

Resources Research 13, 395-403. 
Comerford, N.B., Kidder, G. and Mollitor, A.V. (1984) Importance of subsoil 

fertility to forest and non-forest plant nutrition. In: Stone, E.L. (ed.) Forest 

Soils and Treatment Impacts. University of Tennessee, Knoxville, Tennessee, pp. 

Comfort, S.D., Dick, R.P. and Baham,J. (1991) Air-drying and pretreatment effects 

on soil sulfate sorption. Soil Science Society of America Journal 55, 968-973. 
Condron, L.M., Davis, M.R., Newman, R.H. and Cornforth, I.S. (1996) Influence 

of conifers on the froms of phosphorus in selected New Zealand grassland 

soils. Biology and Fertility of Soils 21, 37-42. 
Conger, B.V. (ed.) (1999) Special issue: allelopathy I. Critical Reviews in Plant Science 

18(5), 577-693. 
Conti, M.E., Palma, R.M., Arrigo, N. and Giardina, E. (1992) Seasonal variations 

of the light organic fractions in soils under different agricultural management 

systems. Communications in Soil Science and Plant Analysis 23, 1693-1704. 
Cooper, P.J.M. (1997) Setting objectives in agroforestry research design. Training 

note. Available at: 

Cooper, P.J.M., Leakey, R.R.B., Rao, M.R. and Reynolds, L. (1996) Agroforestry 

and the mitigation of land degradation in the humid and sub-humid tropics 

of Africa. Experimental Agriculture 32, 235-290. 
Cornejo, F.H., Varela, A. and Wright, S.J. (1994) Tropical forest litter 

decomposition under seasonal drought: nutrient release, fungi and bacteria. 

Oikos 70, 183-190. 
Cornet, F. and Diem, H.G. (1982) Etude comparative de l'efficacite des souches 

de Rhizobium AAcacia isolees de sols du Senegal et effet de la double symbiose 

Rhizobium-Glomus mosseae sur la croissance de Acacia holosericea et A raddiana. 

Bois et Forets des Tropiques 198, 3-15. 
Cornish, P.S. (1993) Soil macrostructure and root growth of establishing seedlings. 

Plant and Soil 151, 119-126. 
Costanza, R., DArge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., 

References 359 

Limburg, K., Naeem, S., O'Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P. 

and van den Belt, M. (1997) The value of the world's ecosystem services and 

natural capital. Nature 387, 253-260. 
Coxson, D.S. (1991) Nutrient release from epiphytic bryophytes in tropical 

montane rain forest (Guadeloupe). Canadian Journal of Botany 69, 2 122-2 129. 
Cressie, N. (1991) Statistics for Spatial Data . John Wiley & Sons, New York, 900 pp. 
Cresswell, H.P. and Kirkegaard, J.A. (1995) Subsoil amelioration by plant roots - 

the process and the evidence. Australian Journal of Soil Research 33, 221-239. 
Crepin, J. and Johnson, R.L. (1993) Soil sampling for environmental assessment. 

In: Carter, M.R. (ed.) Soil Sampling and Methods of Analysis. Lewis Publishers, 

Boca Raton, Florida, pp. 5-18. 
Critchley, W.R.S., Reij, C. and Willcocks, T.J. (1994) Indigenous soil and water 

conservation: a review of the state of knowledge and prospects for building 

on traditions. Land Degradation and Rehabilitation 5, 293-3 14. 
Cross, A.F. and Schlesinger, W.H. (1995) A literature review and evaluation of the 

Hedley fractionation: applications to the biogeochemical cycle of soil 

phosphorus in natural ecosystems. Geoderma 64, 197-214. 
Cuenca, C, Aranguren, J. and Herrera, R. (1983) Root growth and litter 

decomposition in a coffee plantation under shade trees. Plant and Soil 71, 

Cuesta, M., Carlson, G. and Lutz, E. (1994) An Empirical Assessment of Farmers' 

Discount Rates in Costa Rica and its Implication for Soil Conservation. World Bank, 

Environment Department, Washington, DC. 
Cuevas, E. and Medina, E. (1986) Nutrient dynamics within amazonian forest 

ecosystems 1. Nutrient flux in fine litter fall and efficiency of nutrient 

utilization. Oecologia 68, 466-472. 
Cuevas, E. and Medina, E. (1988) Nutrient dynamics within amazonian forests 2. 

Fine root growth, nutrient availability and leaf litter decomposition. Oecologia 

Gumming, J. R. and Weinstein, L.H. (1990) Aluminium-mycorrhizal interactions 

in the physiology of pitch pine seedlings. Plant and Soil 125, 7-18. 
Current, D., Lutz, E. and Scherr, S.J. (1995) The costs and benefits of agroforestry 

to farmers. The World Bank Research Observer 10, 151-180. 
Cushman, J.H. (1984) Nutrient transport inside and outside the root rhizosphere 

- generalized model. Soil Science 138, 164-171. 
Czarnes, S., Hallett, P.D., Bengough, A.G. and Young, I.M. (2000) Root- and 

microbial-derived mucilages affect soil structure and water transport. European 

Journal of Soil Science 51, 435-443. 
Dalton, F.N. (1992) Development of time-domain reflectometry for measuring soil 

water content and bulk soil electrical conductivity. In: Topp, G.C., Reynolds, 

W.D. and Green, R.E. (eds) Advances in Measurement of Soil Physical Properties: 

Bringing Theory into Practice. Soil Science Society of America, Madison, 

Wisconsin, pp. 143-167. 
Damgaard, L.R., Revsbech, N.P. and Reichardt, W. (1998) Use of an oxygen- 
sensitive microscale biosensor for methane to measure methane concentration 

profiles in a rice paddy. Applied and Environmental Microbiology 64, 864-870. 
Dangerfield, J.M. (1990) Abundance, biomass and diversity of soil macrofauna in 

savanna woodland and associated managed habitats. Pedobiologia 34, 141-150. 

360 Refer. 

Darlington, J. P. E.C. (1982) The underground passages and storage pits used in 

foraging by a nest of the termite Macrotermes michaelseni in Kajiado, Kenya. 

Kenya Journal of Zoology 198, 237-247. 
Darrah, P.R. (1991a) Measuring the diffusion-coefficients or rhizosphere exudates 

in soil. 2. The diffusion of sorbing compounds. Journal of Soil Science 42, 42 1-434. 
Darrah, P.R. (1991b) Models of the rhizosphere. 2. A quasi 3-dimensional 

simulation of the microbial-population dynamics around a growing root 

releasing soluble exudates. Plant and Soil 138, 147-158. 
Dash, M.C., Senapati, B.K. and Mishra, C.C. (1980) Nematode feeding by tropical 

earthworms. Oikos 34, 322-325. 
Davies, FT. Jr, Potter, J. R. and Linderman, R.G. (1992) Mycorrhiza and repeated 

drought exposure affect drought resistance and extraradical hyphae 

development of pepper plants independent of plant size and nutrient content. 

Journal of Plant Physiology 139, 289-294. 
Dawson, T.E. (1993) Hydraulic lift and water use by plants: implications for water 

balance, performance and plant-plant interactions. Oecologia 95, 565-574. 
Dean, T.J., Bell, J. P. and Baty, A.J.B. (1987) Soil moisture measurement by an 

improved capacitance technique. I. Sensor design and performance. Journal 

of Hydrology 56,1341-1345. 
Decaens, T. (2000) Degradation dynamics of surface earthworm casts in grasslands 

of the eastern plains of Colombia. Biology and Fertility of Soils 32, 149-156. 
Decaens, T., Lavelle, P., Jimenez, J.J. , Escobar, G. and Rippstein, G. (1994) Impact 

of land management on soil macrofauna in the Oriental Llanos of Colombia. 

European Journal of Soil Biology 30, 157-168. 
Decaens, T., Jimenez, J.J. and Lavelle, P. (1999) Effect of exclusion of the anecic 

earthworm Martiodrilus carimaguensis Jimenez and Moreno on soil properties 

and plant growth in grasslands of the eastern plains of Colombia. Pedobiologia 

Decaens, T., Galvin, J.H. and Amezquita, E. (2001) Proprietes des structures 

produites par les ingenieurs ecologiques a la surface du sol d'une savane 

colombienne. Comte Rendus de I'Academie de Sciences, Life Sciences 324, 465-478. 
Deleporte, S., Hallaire, V. and Tillier, P. (1997) Application of image analysis to a 

quantitative micromorphological study of forest humus. European Journal of 

Soil Biology 33,83-88. 
Delve, R.J., Cadisch, G., Tanner, J. C, Thorpe, W., Thorne, P.J. and Giller, K.E. 

(2001) Implications of livestock feeding management on soil fertility in the 

smallholder farming systems of sub-Saharan Africa. Agriculture, Ecosystems and 

Environment 84, 227-243. 
Desaeger, J. and Rao, M.R. (1999) The root-knot nematode problem in sesbania 

fallows and scope for managing it in western Kenya. Agroforestry Systems 47, 

Deutsch, C.V. and Journel, A. (1992) GSLIB - Geostatistical Software Library and User's 

Guide. Oxford University Press, New York, 341 pp. 
Dewees, PA. (1995) Trees on farms in Malawi: private investment, public policy, 

and farmer choice. World Development 23, 1085-1 102. 
Deweger, LA., Dunbar, P., Mahafee, W.F., Lugtenberg, B.J.J, and Sayler, G.S. 

(1991) Use of bioluminescence markers to detect Pseudomonas spp. in the 

rhizosphere. Applied and Environmental Microbiology 57, 3641-3644. 

References 361 

Di, H.J., Trangmar, B.B. and Kemp, R.A. (1989) Use of geostatistics in designing 
sampling strategies for soil survey. Soil Science Society of America Journal 53, 

Di, H.J., Condron, L.M. and Frossard, E. (1997) Isotope techniques to study 
phosphorus cycling in agricultural and forest soils: a review. Biology and Fertility 
of Soils 24, 1-12. 

Dickson, S. and Kolesik, P. (1999) Visualisation of mycorrhizal fungal structures 
and quantification of their surface area and volume using scanning confocal 
microscopy. Mycorrhiza 9, 205-213. 

Dickson, S. and Smith, S.E. (1998) Evaluation of vesicular-arbuscular mycorrhizal 
colonisation by staining. In: Varma, A. (ed.) Mycorrhiza Manual. Springer, 
Berlin, pp. 77-84. 

Diels, J., Aihou, K., Iwuafor, E.N.O., Merckx, R., Lyasse, O., Sanginga, N., 
Vanlauwe, B. and Deckers, J. (2002) Options for soil organic carbon 
maintenance under intensive cropping in the West-African savanna. In: 
Vanlauwe, B., Diels, J., Sanginga, N. and Merckx, R. (eds) Integrated Nutrient 
Management in Sub-Saharan Africa: from Concept to Practice. CAB International, 
Wallingford, UK, pp. 299-312. 

DiGiovanni, G.D., Watrud, L.S., Seidler, R.J. and Widmer, F. (1999) Comparison 
of parental and transgenic alfalfa rhizosphere bacterial communities using 
Biolog GN metabolic fingerprinting and enterobacterial repetitive intergenic 
consensus sequence-PCR (ERIC-PCR). Microbial Ecology 37, 129-139. 

Dinkelaker, B., Hahn, G., Romheld, V., Wolf, G.A. and Marschner, H. (1993a) 
Nondestructive methods for demonstrating chemical changes in the 
rhizosphere. 1. Description of methods. Plant and Soil 156, 67-70. 

Dinkelaker, B., Hahn, G. and Marschner, H. (1993b) Nondestructive methods for 
demonstrating chemical changes in the rhizosphere. 2. Application of 
methods. Plant and Soil 156, 71-74. 

Dirksen, C. and Dasberg, S. (1993) Improved calibration of time domain 
refiectometry soil water content measurements. Soil Science Society of America 
Journal 57, 660-667. 

Dodd, J.C., Arias, I., Koomen, I. and Hayman, D.S. (1990a) The management of 
populations of vesicular-arbuscular mycorrhizal fungi in acid-infertile soils of 
a savanna ecosystem. I. The effect of pre-cropping and inoculation with VAM 
fungi on plant growth and nutrition in the field. Plant and Soil 122, 229-240. 

Dodd, J. C, Arias, I., Koomen, I. and Hayman, D.S. (1990b) The management of 
populations of vesicular-arbuscular mycorrhizal fungi in acid-infertile soils of 
a savanna ecosystem. II. The effect of pre-crops on the spore populations of 
native and introduced VAM-fungi. Plant and Soil 122, 241-247. 

Dodd, J. C, Rosendahl, S., Giovannetti, M., Broome, A., Lanfranco, L. and Walker, 
C. (1996) Inter- and intraspecific variation within the morphologically similar 
fungi Glomus mosseae and Glomus coronatum. New Phytologist 133, 1 13-122. 

Donald, R.G., Anderson, D.W. and Stewart, J.W.B. (1993) Potential role of 
dissolved organic carbon in phosphorus transport in forested soils. Soil Science 
Society of America Journal 57, 1611-1618. 

Donovan, G. and Casey, F. (1998) Soil Fertility Management in Sub-Saharan Africa. 
World Bank Technical Paper No. 408. The World Bank, Washington, DC, 
61 pp. 

362 Refer. 

Dorrance, D.W., Wilson, L.G., Everett, L.G. and Cullen, S.J. (1991) Compendium 

oiinsitu pore-liquid samplers for vadose zone. In: Nash, R.G. and Leslie, A.R. 

(eds) Groundwater Residue Sampling Designs. American Chemical Society, 

Washington, DC, pp. 300-333. 
Douglas, J. T. (1986) Macroporosity and permeability of some soil cores from 

England and France. Geoderma 37, 221-231. 
Douglas, M.G. (1994) Sustainable Use of Agricultural Soils: a Review of the Prerequisites 

for Success or Failure. Development and Environment Report No. 1 1. University 

of Bern, Bern, 162 pp. 
Doumenge, C, Gilmour, D., Ruiz Perez, M. and Blockhus, J. (1995) Tropical 

montane cloud forests: conservation status and management issues. In: 

Hamilton, L.S., Juvik, J.O. and Scatena, F.N. (eds) Tropical Montane Cloud 

Forests. Springer, Berlin, pp. 24-37. 
Draaijers, G.P.J., Erisman, J.W., Spranger, T. and Wyers, G.P. (1996) The 

application of throughfall measurements for atmospheric deposition 

monitoring. Atmospheric Environment 30, 3349-3361. 
Drechsel, P., Glaser, B. and Zech, W. (1991) Effect of four multipurpose tree species 

on soil amelioration during fallow in central Togo. Agroforestry Systems 16, 

Drechsel, P., Steiner, K.G. and Hagedorn, F. (1996) A review on the potential of 

improved fallows and green manure in Rwanda. Agroforestry Systems 33, 

Dreyfus, B.L. and Dommergues, Y.R. (1981) Nodulation oiAcacia species by fast- 

and slow-growing tropical strains of Rhizobium. Applied and Environmental 

Microbiology 41, 97-99. 
Dubach, M. and Russelle, M.P. (1995) Reducing the cost of estimating root 

turnover with horizontally installed minirhizotrons. Agronomy Journal 87, 

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, PA. and Smith, F. (1956) 

Colorimetric method for determination of sugars and related substances. 

Analytical Chemistry 28, 350-356. 
Duchesne, L.C., Peterson, R.L. and Ellis, B.E. (1989) The time-course of disease 

suppression and antibiosis by the ectomycorrhizal fungus Paxillus involutus. 

New Phytologist 1 1 1, 693-698. 
Dudal, R. and Deckers, J. (1993) Soil organic matter in relation to soil productivity. 

In: Mulongoy, K. and Merckx, R. (eds) Soil Organic Matter Dynamics and 

Sustainability of Tropical Agriculture. John Wiley & Sons, Chichester, UK, 

pp. 377-380. 
Duguma, B., Kang, B.T. and Okali, D.U.U. (1988) Effect of pruning intensities of 

three woody leguminous species grown in alley cropping with maize and 

cowpea on an alfisol. Agroforestry Systems 6, 19-35. 
Dunne, T. (1977) Evaluation of erosion conditions and trends. In: Kunkle, S.H. 

(ed.) Guidelines for Watershed Management. Food and Agriculture Organization 

of the United Nations, Rome, pp. 53-83. 
Duponnois, R., Senghor, K., Thioulouse, J. and Ba, A.M. (1999) Susceptibility of 

several sahelian Acacia to Meloidogyne javanica (Treub) Chitw. Agroforestry 

Systems 46, 123-130. 
Duponnois, R., Founoune, H., Ba, A.M., Plenchette, C, Jaafari, S.E., Neyra, M. 

References 363 

and Ducousso, M. (2000) Ectomycorrhization of Acacia holosericea A. Cunn. ex 

G. Don by Pisolithus spp. in Senegal: effect on plant growth and on the root- 
knot nematode Meloidogyne javanica. Annals of Forest Science 57, 345-350. 
Dupraz, C. (1999) Adequate design of control treatments in long term agroforestry 

experiments with multiple objectives. Agroforestry Systems 43, 35-48. 
Dyke, G.V. (1988) Comparative Experiments with Field Crops, 2nd edn. Griffin, 

London, 262 pp. 
Eamus, D. (1999) Ecophysiological traits of deciduous and evergreen woody species 

in the seasonally dry tropics. Trends in Ecology and Evolution 14, 1 1-16. 
Eastham, J., Rose, C.W., Cameron, D.M., Ranee, S.J. and Talsma, T. (1988) The 

effect of tree spacing on evaporation from an agroforestry experiment. 

Agricultural and Forest Meteorology 42, 355-368. 
Eastham, J., Scott, P.R. and Steckis, R. (1994) Components of the water balance 

for tree species under evaluation for agroforestry to control salinity in the 

wheatbelt of western Australia. Agroforestry Systems 26, 157-169. 
Eberl, L., Givskov, M., Poulsen, L.K. and Molin, S. (1997) Use of bioluminescence 

for monitoring the viability of individual Pseudomonas putida KT2442 cells. 

FEMS Microbiology Letters 149, 133-140. 
Edmunds, W.M. and Gaye, C.B. (1997) Naturally high nitrate concentrations in 

groundwaters from the Sahel. Journal of Environmental Quality 26, 1231-1239. 
Edwards, S.G., Fitter, A.H. and Young, J-P.W. (1997) Quantification of an 

arbuscular mycorrhizal fungus, Glomus mosseae, within plant roots by 

competitive polymerase chain reaction. Mycological Research 101, 1440-1444. 
Egener, T., Hurek, T. and Reinhold-Hurek, B. (1998) Use of green fluorescent 

protein to detect expression of nif genes of Azoarcus sp. BH72, a grass- 
associated diazotroph, on rice roots. Molecular Plant-Microbe Interactions 1 1, 

Egger, K. (1995) Molecular analysis of ectomycorrhizal fungal communities. 

Canadian Journal of Botany 73, S1415-S1422. 
Egger, K.N. and Sigler, L. (1993) Relatedness of the ericoid endophytes Scytalidium 

vaccinii and Hymenoscyphus ericae inferred from analysis of ribosomal DNA. 

Mycologia 85,219-230. 
Egoumenides, C. (1990) Fractions de l'azote organique dans les sols tropicaux et 

fertilite azotee. In: IRAT (ed.) Agronomie et Ressources Naturelles en Regions 

Tropicales. IRAT/CIRAD, Montpellier, France, pp. 317-326. 
Ehrlich, D., Lambin, E.F. and Malingreau, J. -P. (1997) Biomass burning and 

broad-scale land-cover changes in Western Africa. Remote Sensing of 

Environment 61, 201-209. 
Eichert, T., Goldbach, H.E. and Burkhardt, J. (1998) Evidence for the uptake of 

large anions through stomatal pores. Botanica Acta 111, 461-466. 
Eissenstat, D.M., Wells, C.E., Yanai, R.D. and Whitbeck, J.L. (2000) Building roots 

in a changing environment: implications for root longevity. New Phytologist 

Ekanade, O. (1985) The effects of cocoa cultivation on some physical properties of 

soil in south-western Nigeria. The International Tree Crops Journal 3, 113-124. 
Ekanade, O. (1987) Small-scale cocoa farmers and environmental change in the 

tropical rain forest regions of south-western Nigeria. Journal of Environmental 

Management 25, 61-70. 

364 Refer. 

Ekblad, A. and Huss-Danell, K. (1995) Nitrogen fixation by Alnus incana and 

nitrogen transfer from A. incana to Pinus sylvestris influenced by macronutrients 

and ectomycorrhiza. New Phytologist 131, 453-459. 
Eldredge, E.P., Shock, C.C. and Stieber, T.D. (1993) Calibration of granular matrix 

sensors for irrigation management. Agronomy Journal 85, 1228-1232. 
Elmes, R.P. and Mosse, B. (1984) Vesicular-arbuscular endomycorrhizal inoculum 

production. II. Experiments with maize (Zea mays) and other hosts in nutrient 

flow culture. Canadian Journal of Botany 62, 1531-1536. 
Elrick, D.E. and Reynolds, W.D. (1992) Infiltration from constant-head well 

permeameters and infiltrometers. In: Topp, G.C., Reynolds, W.D. and Green, 

R.E. (eds) Advances in Measurement of Soil Physical Properties: Bringing Theory into 

Practice. Soil Science Society of America, Madison, Wisconsin, pp. 1-24. 
Eltrop, L. (1993) Role of Ectomycorrhiza in the mineral nutrition of Norway spruce 

seedlings (Picea abies (L.) Karst.). PhD thesis, University of Hohenheim, 

Hohenheim, 166 pp. 
Emerson, WW. and Greenland, D.J. (1990) Soil aggregate-formation and stability. 

In: de Boots, M., Hayes, M. and Herbillon, A. (eds) Soil Colloids and their 

Associations in Aggregates. Plenum Press, New York, pp. 485-511. 
Ephrath, J.E., Silberbush, M. and Berliner, P.R. (1999) Calibration of 

minirhizotron readings against root length density data obtained from soil 

cores. Plant and Soil 209, 201-208. 
Eriksen, J., Lefroy, R.D.B. and Blair, G.J. (1995) Physical protection of soil organic 

S studied using acetylacetone extraction at various intensities of ultrasonic 

dispersion. Soil Biology and Biochemistry 27, 1005-1010. 
Eriksen, J., Murphy, M.D. and Schnug, E. (1998) The soil sulphur cycle. In: 

Schnug, E. (ed.) Sulphur in Agroecosystems. Kluwer Academic Publishers, 

Dordrecht, pp. 39-73. 
Erland, S. (1995) Abundance of Tylospora fibrillosa ectomycorrhizas in a South 

Swedish spruce forest measured by RFLP analysis of the PCR-amplified rDNA 

ITS region. Mycological Research 99, 1425-1428. 
Eschenbrenner, V. (1986) Contribution des termites a la microagregation des sols 

tropicaux. Cahiers ORSTOM, serie Pedologie 22, 397-408. 
Estrella, M., Blauert,J., Campilan, D., Gaventa,J., Gonsalves, J., Guijt, I., Johnson, 

D. and Ricafort, R. (2000) Learning from Change: Issues and Experiences in 

Participatory Monitoring and Evaluation. Intermediate Technology Publications, 

London, 274 pp. 
Evans, D.O. and Szott, L.T. (1995) Nitrogen Fixing Trees for Acid Soils. Winrock 

International Institute for Agricultural Development, Morrilton, Arkansas, 

328 pp. 
Evett, S.R. and Steiner, J.L. (1995) Precision of neutron scattering and capacitance 

type soil water content gauges from field calibration. Soil Science Society of 

America Journal 59, 961-968. 
Ewel, J.J. (1986) Designing agricultural ecosystems for the humid tropics. Annual 

Reviews in Ecology and Systematics 17, 245-271. 
Faber, B.A., Zasoski, R.J., Munns, D.N. and Shackel, K. (1991) A method of 

measuring hyphal nutrient and water uptake in mycorrhizal plants. Canadian 

Journal of Botany 69, 87-94. 
Fabiao, A., Persson, HA. and Steen, E. (1985) Growth dynamics of superficial roots 

References 365 

in Portugese plantations of Eucalyptus globulus Labill. studied with a mesh bag 

technique. Plant and Soil 83, 233-242. 
Fairhurst, T., Lefroy, R., Mutert, E. and Batjes, N. (1999) The importance, 

distribution and causes of phosphorus deficiency as a constraint to crop 

production in the tropics. Agroforestry Forum 9, 2-7. 
Fairley, R.I. and Alexander, I.J. (1985) Methods of calculating fine root production 

in forests. In: Fitter, A.H., Atkinson, D., Read, D.J. and Usher, M.B. (eds) 

Ecological Interactions in Soil. Blackwell Scientific Publications, Oxford, 

pp. 37-42. 
FAO (1976) Conservation in Arid and Semi-arid Zones. Food and Agriculture 

Organization of the United Nations, Rome, 125 pp. 
Fardeau, J.-O, Guiraud, G. and Marol, C. (1996) The role of isotope techniques 

on the evaluation of the agronomic effectiveness of P fertilizers. Fertilizer 

Research 45, 101-109. 
de Faria, S.M., Lewis, G.P., Sprent, J.I. and Sutherland, J. M. (1989) Occurrence 

of nodulation in the Leguminosae. New Phytologist 111, 607-619. 
Farley, RA. and Fitter, A.H. (1999) Temporal and spatial variation in soil resources 

in a deciduous woodland. Journal of Ecology 87, 688-696. 
Federer, W.T. (1993) Statistical Design and Analysis of Intercropping Experiments. Vol. 

1: Two Crops. Springer, New York, 304 pp. 
Federer, W.T. (1998) Statistical Design and Analysis of Intercropping Experiments . Vol. 

2: Three or More Crops. Springer, New York, 293 pp. 
Feigl, B.J., Melillo, J. and Cerri, C.C. (1995) Changes in the origin and quality of 

soil organic matter after pasture introduction in Rondonia (Brazil). Plant and 

Soil 175,21-29. 
Feldmann, F. and Idczak, E. (1994) Inoculum production of vesicular-arbuscular 

mycorrhizal fungi for use in tropical nurseries. In: Norris, J.R., Read, D. and 

Varma, A.K. (eds) Techniques for Mycorrhizal Research. Academic Press, London, 

pp. 799-817. 
Feller, C. and Beare, M.H. (1997) Physical control of soil organic matter dynamics 

in the tropics. Geoderma 79, 69-1 16. 
Feller, C, Francois, C, Villemin, G., Portal, J. -M., Toutain, F. and Morel, J. -L. 

(1991a) Nature des matieres organiques associees aux fractions argileuses d'un 

sol ferralitique. Comptes Rendus de VAcademie des Sciences, Paris 312, 1491-1497. 
Feller, C, Fritsch, E., Poss, R. and Valentin, C. (1991b) Effet de la texture sur le 

stockage et la dynamique des matieres organiques dans quelques sols 

ferrugineux et ferralitiques (Afrique de l'Ouest, en particulier). Cahiers 

ORSTOM, serie Pedologie 26, 25-36. 
Feller, C, Burtin, G., Gerard, B. and Balesdent, J. (1991c) Utilisation des resines 

sodiques et des ultrasons dans le fractionnement granulometrique de la 

matiere organique des sols. Interet et limites. Science du Sol 29, 77-93. 
Fernandez, J. E., Clothier, B.E. and van Noordwijk, M. (2000) Water uptake. In: 

Smit, A.L., Bengough, A.G., Engels, C, van Noordwijk, M., Pellerin, S. and 

van der Geijn, S.C. (eds) Root Methods: a Handbook. Springer, Berlin, pp. 

Ferrier, R.C. and Alexander, I.J. (1991) Internal redistribution of N in Sitka spruce 

seedlings with partly droughted root systems. Forest Science 37, 860-870. 
Fielding, W.J. and Sherchan, D.P. (1999) The variability of level and sloping 

366 Refer 

terraces in Eastern Nepal and the implications for the design of experiments. 

Experimental Agriculture 35, 449-459. 
Fies, J.C. and Bruand, A. (1990) Textural porosity analysis of a silty clay soil using 

pore volume balance estimation, mercury porosimetry and quantified 

backscattered electron scanning images (BESI). Geoderma 47, 209-219. 
Finck, A. (1992) Dunger und Dungung . VCH Verlagsgesellschaft, Weinheim, 488 pp. 
Finkelstein, P.L. and Sims, P.F. (2001) Sampling error in eddy correlation flux 

measurements. Journal of Geophysical Research 106, 3503-3509. 
Finlay, R.D. and Read, D.J. (1986a) The structure and function of the vegetative 

mycelium of ectomycorrhizal plants. I. Translocation of 14 C-labelled carbon 

between plants interconnected by a common mycelium. New Phytologist 103, 

Finlay, R.D. and Read, D.J. (1986b) The structure and function of the vegetative 

mycelium interconnecting plants. II. The uptake and distribution of 

phosphorus by mycelium interconnecting host plants. New Phytologist 103, 

Finzi, A.C., Canham, CD. and van Breemen, N. (1998) Canopy tree-soil 

interactions within temperate forests: species effects on pH and cations. 

Ecological Applications 8, 447-454. 
Fisher, R.F. (1995) Amelioration of degraded rain forest soils by plantations of 

native trees. Soil Science Society of America Journal 59, 544-549. 
Fisher, R.F. and Juo, A.S.R. (1995) Mechanisms of tree growth in acid soils. In: 

Evans, D.O. and Szott, L.T. (eds) Nitrogen Fixing Trees for Acid Soils. Winrock 

International Institute for Agricultural Development, Morrilton, Arkansas, 

pp. 1-18. 
Fitter, A.H. (1986) Spacial and temporal patterns of root activity in a species-rich 

alluvial grassland. Oecologia 69, 594-599. 
Fitter, A.H. (1988) Water relations of red clover Trifolium pratense L. as affected by 

VA mycorrhizal infection and phosphorus supply before and during drought. 

Journal of Experimental Botany 39, 595-603. 
Fitzpatrick, E.A. (1990) Roots in thin section of soils. Development in Soil Science 19, 

Fixen, P.E. and Grove, J. H. (1990) Testing soils for phosphorus. In: Westerman, 

R.L. (ed.) Soil Testing and Plant Analysis. Soil Science Society of America, 

Madison, Wisconsin, pp. 141-180. 
Fogel, R. and Hunt, G. (1983) Contribution of mycorrhizae and soil fungi to 

nutrient cycling in a Douglas-fir ecosystem. Canadian Journal of Forest Research 

Fox, C.A., Guertin, R.K., Dickson, E., Sweeney, S., Protz, R. and Mermut, A.R. 

(1993) Micromorphological methodology for inorganic soils. In: Carter, M.R. 

(ed.) Soil Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, 

Florida, pp. 683-709. 
Fox, R.L. and Kamprath, E.J. (1970) Phosphate sorption isotherms for evaluating 

the phosphate requirements of soils. Soil Science Society of America Proceedings 

34, 902-907. 
Fox, R.L. and Lipps, R.C. (1964) A comparison of stable strontium and P32 as 

tracers for estimating alfalfa root activity. Plant and Soil 20, 337-350. 
Fragoso, C. and Rojas, P. (1994) Soil biodiversity and land management in the 

References 367 

tropics - the case of ants and earthworms. In: Transactions of the 15th World 

Congress of Soil Science, Acapulco, Mexico, July 1994, Vol. 4a. International Soil 

Science Society, Vienna, pp. 232-237. 
Fragoso, C, Kanyonyo ka Kajondo,J., Blanchart, E., Moreno, A.G., Senapati, B.K. 

and Rodriguez, C. (1999) A survey of tropical earthworms: taxonomy, 

biogeography and environmental plasticity. In: Lavelle, P., Brussaard, L. and 

Hendrix, P. (eds) Earthworm Management in Tropical Agroecosystems. CAB 

International, Wallingford, UK, pp. 1-27. 
Francis, R. and Read, D.J. (1984) Direct transfer of carbon between plants 

connected by vesicular-arbuscular mycorrhizal mycelium. Nature 307, 53-56. 
Franken, G., Pijpers, M. and Matzner, E. (1995) Aluminium chemistry of the soil 

solution in an acid forest soil as influenced by percolation rate and soil 

structure. European Journal of Soil Science 46, 613-619. 
Franzel, S., Coe, R., Cooper, P.J., Place, F. and Scherr, S.J. (2001) Assessing the 

adoption potential of agroforestry practices in sub-Saharan Africa. Agricultural 

Systems 69, 37-62. 
Franzluebbers, A.J., Haney, R.L., Hons, F.M. and Zuberer, DA. (1996) Active 

fraction of organic matter in soils with different texture. Soil Biology and 

Biochemistry 28, 1367-1372. 
Freney, J.R. (1986) Forms and reactions of organic sulfur compounds in soil. In: 

Tabatabai, M.A. (ed.) Sulfur in Agriculture. American Society of Agronomy, 

Madison, Wisconsin, pp. 207-232. 
Freney, J.R. , Melville, G.E. and Williams, C.H. (1971) Organic sulphur fractions 

labelled by additions of 35 S-sulphate to soil. Soil Biology and Biochemistry 3, 

Frey, B. and Schiiepp, H. (1993) A role of vesicular-arbuscular (VA) mycorrhizal 

fungi in facilitating inter-plant nitrogen transfer. Soil Biology and Biochemistry 

Friesen, D.K., Rao, I.M., Thomas, R.J., Oberson, A. and Sanz, J.I. (1997) 

Phosphorus acquisition and cycling in crop and pasture systems in low fertility 

soils. Plant and Soil 196, 289-294. 
Frossard, E. and Sinaj, S. (1997) The isotope exchange kinetic technique: a method 

to describe the availability of inorganic nutrients. Applications to K, P0 4 , S0 4 

and Zn. Isotopes in Environmental and Health Studies 33, 61-77. 
Fujisaka, S. (1993) A case of farmer adaptation and adoption of contour hedgerows 

for soil conservation. Experimental Agriculture 29, 97-105. 
Gage, D.J., Bobo, T. and Long, S.R. (1996) Use of green fluorescent protein to 

visualize the early events of symbiosis between Rhizobium meliloti and alfalfa 

(Medicago sativa). Journal of Bacteriology 178, 7159-7166. 
Gahoonia, T.S., Raza, S. and Nielsen, N.E. (1994) Phosphorus depletion in the 

rhizosphere as influenced by soil moisture. Plant and Soil 159, 213-218. 
Gardner, R. (1997) The third dimension: soil fertility and integrated nutrient 

management on hillsides. In: Gregory, P.J., Pilbeam, C.J. and Walker, S.H. 

(eds) Integrated Nutrient Management on Farmers' Fields: Approaches that Work. 

The University of Reading, Reading, pp. 173-192. 
Gardner, R., Mawdesley, K., Tripathi, B., Gaskin, S. and Adams, S. (2000) Soil 

Erosion and Nutrient Loss in the Middle Hills of Nepal (1996-1998). Queen Mary 

and Westfield College, University of London, London, 57 pp. 

368 Refer. 

Gardner, W.H. (1986) Water content. In: Klute, A. (ed.) Methods of Soil Analysis, 

Part 1. American Society of Agronomy, Madison, Wisconsin, pp. 479-492. 
Garg, V.K. and Jain, R.K. (1992) Influence of fuelwood trees on sodic soils. 

Canadian Journal of Forest Research 22, 729-735. 
Garrity, D.P. (1996) Tree-soil-crop interactions on slopes. In: Ong, C.K. and 

Huxley, P. (eds) Tree-Crop Interactions. CAB International, Wallingford, UK, 

pp. 299-318. 
Garrity, D.P., Soekardi, M., van Noordwijk, M., de la Cruz, R., Pathak, P.S., 

Gunasena, H.P.M., van So, N., Huijun, G. and Majid, N.M. (1997) The 

Imperata grasslands of tropical Asia: area, distribution, and typology. 

Agroforestry Systems 36, 3-29. 
Gass, W.B., Peterson, G.A., Hauck, R.D. and Olson, R.A. (1971) Recovery of 

residual nitrogen by corn (Zea mays L.) from various soil depths as measured 

by 15 N tracer techniques. Soil Science Society of America Proceedings 35, 

Gaunt, J.L., Sohi, S.P., Arah, J.R.M., Mahieu, N., Yang, H. and Powlson, D.S. 

(2000) A physical soil fractionation scheme to obtain organic matter fractions 

suitable for modelling. In: Kirk, G.J.D. and Oik, D.C. (eds) Carbon and Nitrogen 

Dynamics in Flooded Soils. International Rice Research Institute, Los Banos, 

Philippines, pp. 89-99. 
Geelhoed, J.S., VanRiemsdijk, W.H. and Findenegg, G.R. (1999) Simulation of 

the effect of citrate exudation from roots on the plant availability of phosphate 

adsorbed on goethite. European Journal of Soil Science 50, 379-390. 
van Genuchten, M.T. (1980) A closed-form equation for predicting the hydraulic 

conductivity of unsaturated soils. Soil Science Society of America Journal 44, 

George, E., Haussler, K.U., Vetterlein, D., Gorgus, E. and Marschner, H. (1992) 

Water and nutrient translocation by hyphae of Glomus mosseae. Canadian 

Journal of Botany 70, 2130-2137. 
George, E., Marschner, H. and Jakobsen, I. (1995) Role of arbuscular mycorrhizal 

fungi in uptake of phosphorus and nitrogen from soil. Critical Reviews in 

Biotechnology 15, 257-270. 
Germida, J.J. and Siciliano, S.D. (1998) Biolog analysis and fatty acid methyl ester 

profiles indicate that pseudomonad inoculants that promote phytore- 

mediation alter the root-associated microbial community of Bromus biebersteinii. 

Soil Biology and Biochemistry 30, 1717-1723. 
Ghani, A., McLaren, R.G. and Swift, R.S. (1992) Sulfur mineralization and 

transformations in soils as influenced by addition of carbon, nitrogen and 

sulfur. Soil Biology and Biochemistry 24, 33 1-341. 
Ghani, A., McLaren, R.G. and Swift, R.S. (1993) The incorporation and 

transformation of 3o S in soil: effects of soil conditioning and glucose or sulphate 

additions. Soil Biology and Biochemistry 25, 327-335. 
Ghodrati, M. and Jury, W.A. (1990) A field study using dyes to characterize 

preferential flow of water. Soil Science Society of America Journal 54, 1558-1563. 
Gianello, C. and Bremner, J.M. (1986) A simple chemical method of assessing 

potentially available organic nitrogen in soil. Communications in Soil Science and 

Plant Analysis 17, 195-214. 
Gijsman, A.J., Floris, J., van Noordwijk, M. and Brouwer, G. (1991) An inflatable 

References 369 

minirhizotron system for root observations with improved soil/tube contact. 

Plant and Soil 134, 261-269. 
Gill, R.A. and Jackson, R.B. (2000) Global patterns of root turnover for terrestrial 

ecosystems. New Phytologist 147, 13-31. 
Giller, K.E. (1997) Tropical legumes: providers and plunderers of nitrogen. In: 

Bergstrom, L. and Kirchmann, H. (eds) Carbon and Nitrogen Dynamics in Natural 

and Agricultural Tropical Ecosystems. CAB International, Wallingford, UK, pp. 

Giller, K.E. (2000) Translating science into action for agricultural development in 

the tropics: an example from decomposition studies. Applied Soil Ecology 14, 

Giller, K.E. (2001) Nitrogen Fixation in Tropical Cropping Systems, 2nd edn. CAB 

International, Wallingford, UK, 448 pp. 
Giller, K.E. and Cadisch, G. (1995) Future benefits from biological nitrogen 

fixation: an ecological approach to agriculture. Plant and Soil 174, 255-277. 
Giller, K.E. and Day, J.M. (1985) Nitrogen fixation in the rhizosphere: significance 

in natural and agricultural ecosystems. In: Fitter, A.H. (ed.) Ecological 

Interactions in Soil: Plants, Microbes and Animals. Blackwell Scientific, Oxford, 

pp. 127-147. 
Giller, K.E., Ormesher, J. and Awah, F.M. (1991) Nitrogen transfer from Phaseolus 

bean to intercropped maize measured using lo N-enrichment and lo N-isotope 

dilution methods. Soil Biology and Biochemistry 23, 339-346. 
Giller, K.E., Cadisch, G., Ehaliotis, C, Adams, E., Sakala, W.D. and Mafongoya, 

P.L. (1997) Building soil nitrogen capital in Africa. In: Buresh, R.J., Sanchez, 

P.A. and Calhoun, F. (eds) Replenishing Soil Fertility in Africa. Soil Science 

Society of America, Madison, Wisconsin, pp. 151-192. 
Giller, K.E., Witter, E. and McGrath, S.P. (1998) Toxicity of heavy metals to 

microorganisms and microbial processes in agricultural soils - a review. Soil 

Biology and Biochemistry 30, 1389-1414. 
Gilot, C, Lavelle, P., Blanchart, E., Keli, J., Kouassi, P. and Gulliaume, G. (1995) 

Biological activity of soil under rubber plantations in Cote d'lvoire. Acta 

Zoologica Fennica 196, 186-190. 
Giovannetti, M. and Mosse, B. (1980) An evaluation of techniques for measuring 

vesicular arbuscular mycorrhizal infection in roots. New Phytologist 84, 

Gittinger, J.P. (1982) Economic Analysis of Agricultural Projects, 2nd edn. The Johns 

Hopkins University Press, Baltimore, Maryland, 505 pp. 
Glavac, V., Koenies, H. and Ebben, U. (1990) Seasonal variations in mineral 

concentrations in the trunk xylem sap of beech in a 42 year old beech forest 

stand. New Phytologist 1 16, 47-54. 
Gobel, C, Hahn, A., Giersch, T. and Hock, B. (1998) Monoclonal antibodies for 

the identification of arbuscular mycorrhizal fungi. In: Varma, A. (ed.) 

Mycorrhiza Manual. Springer, Berlin, pp. 271-288. 
Golchin, A., Oades, J.M., Skjemstad, J.O. and Clarke, P. (1995) Structural and 

dynamic properties of soil organic matter as reflected by 13 C natural 

abundance, pyrolysis mass spectrometry and solid-state 13 C NMR spectroscopy 

in density fractions of an oxisol under forest and pasture . Australian Journal of 

Soil Research 33, 59-76. 

370 Refer. 

Golden Software (1999) Surfer 7, User's Guide. Golden Software, Colorado, 619 pp. 
Goovaerts, P. (1998) Geostatistical tools for characterizing the spatial variability of 

microbial and physico-chemical soil properties. Biology and Fertility of Soils 27, 

Gordon, W.S. and Jackson, R.B. (2000) Nutrient concentrations in fine roots. 

Ecology 81, 275-280. 
Gorres, J.H., Dichiaro, M.J., Lyons, J.B. and Amador, J.A. (1998) Spatial and 

temporal patterns of soil biological activity in a forest and an old field. Soil 

Biology and Biochemistry 30, 219-230. 
Gottlein, A., Hell, U. and Blasek, R. (1996) A system for microscale tensiometry 

and lysimetry. Geoderma 69, 147-156. 
Gottwald, T.R., Avinent, L., Llacer, G., Hermoso de Mendoza, A. and Cambra, M. 

(1995) Analysis of the spatial spread of sharka (plum pox virus) in apricot and 

peach orchards in eastern Spain. Plant Disease 79, 266-278. 
Gouyon, A., de Foresta, H. and Levang, P. (1993) Does jungle rubber' deserve its 

name? An analysis of rubber agroforestry systems in southeastern Sumatra. 

Agroforestry Systems 22, 181-206. 
Govindarajan, M., Rao, M.R., Mathuva, M.N. and Nair, P.K.R. (1996) Soil-water 

and root dynamics under hedgerow intercropping in semiarid Kenya. 

Agronomy Journal 88, 513-520. 
Graham, J.H. and Sylvertson, J. P. (1984) Influence of vesicular arbuscular 

mycorrhiza on the hydraulic conductivity of roots of two citrus rootstocks. New 

Phytologist 97, 277-284. 
Granli, T. and Bockman, O.C. (1994) Nitrous oxide from agriculture. Norwegian 

Journal of Agricultural Sciences Supplement 12, 1-128. 
Grasse, P.P. (1984) Termitologia . Masson, Paris, 613 pp. 
Grayston, S.J. and Campbell, CD. (1996) Functional biodiversity of microbial 

communities in the rhizospheres of hybrid larch (Larix eurolepis) and Sitka 

spruce (Pic ea site hensis). Tree Physiology 16, 1031-1038. 
Grayston, S.J., Vaughan, D. and Jones, D. (1997) Rhizosphere carbon flow in trees, 

in comparison with annual plants: the importance of root exudation and its 

impact on microbial activity and nutrient availability. Applied Soil Ecology 5, 

Greacen, E.L. (1981) Soil Water Assessment by the Neutron Method. CSIRO, Victoria, 

Australia, 140 pp. 
Green, R.E., Ahuja, L.R. and Chong, S.K. (1986) Hydraulic conductivity, 

diffusivity, and sorptivity of unsaturated soils: field methods. In: Klute, A. (ed.) 

Methods of Soil Analysis, Part 1. American Society of Agronomy, Madison, 

Wisconsin, pp. 771-798. 
Green, S.R., Grace, J. and Hutchings, N.J. (1995) Observations of turbulent air- 
flow in 3 stands of widely spaced sitka spruce . Agricultural and Forest Meteorology 

74, 205-225. 
Greenland, D.J. (1979) Structural organization of soils and crop production. In: 

Lai, R. and Greenland, D.J. (eds) Soil Physical Properties and Crop Production in 

the Tropics. John Wiley & Sons, Chichester, UK, pp. 47-56. 
Greenland, D.J. (1994) Long-term cropping experiments in developing countries. 

In: Leigh, RA. and Johnston, A.E. (eds) Long-term Experiments in Agricultural 

and Ecological Sciences. CAB International, Wallingford, UK, pp. 187-209. 

References 371 

Gregorich, E.G. and Ellert, B.H. (1993) Light fraction and macroorganic matter 

in mineral soils. In: Carter, M.R. (ed.) Soil Sampling and Methods of Analysis. 

Lewis Publishers, Boca Raton, Florida, pp. 397-407. 
Gregorich, E.G., Carter, M.R., Angers, D.A., Monreal, CM. and Ellert, B.H. (1994) 

Towards a minimum data set to assess soil organic matter quality in 

agricultural soils. Canadian Journal of Soil Science 74, 367-385. 
Gregory, P.J. (1988) Water and crop growth. In: Wild, A. (ed.) Russell's Soil 

Conditions and Plant Growth. Longman, Harlow, pp. 338-377. 
Greminger, P.J., Sud, Y.K. and Nielsen, D.R. (1985) Spatial variability of field 

measured soil water characteristics. Soil Science Society of America Journal 49, 

Grevers, M.C.J, and Jong, E.D. (1994) Evaluation of soil-pore continuity using 

geostatistical analysis on macroporosity in serial sections obtained by computed 

tomography scanning. In: Anderson, S.H. and Hopmans, J.W. (eds) 

Tomography of Soil-Water-Root Processes. Soil Science Society of America, 

Madison, Wisconsin, pp. 73-86. 
Griffith, D.M. (2000) Agroforestry: a refuge for tropical biodiversity after fire. 

Conservation Biology 14, 325-326. 
Griffiths, B.S., Ritz, K., Ebblewhite, N. and Dobson, G. (1999) Soil microbial 

community structure: effects of substrate loading rates. Soil Biology and 

Biochemistry 31, 145-153. 
Grimaldi, C, Grimaldi, M. and Vauclin, M. (1994) The effect of the chemical 

composition of a ferrallitic soil on neutron probe calibration. Soil Technology 

7, 233-247. 
Grimaldi, M., Sarrazin, M., Chauvel, A., Luizao, F.J., Nunes, N., Rodrigues, 

M.R.L., Amblard, P. and Tessier, D. (1993) Effets de la deforestation et des 

cultures sur la structure des sols argileux d'Amazonie bresilienne. Cahiers 

Agriculture 2, 36-47. 
Grist, P., Menz, K. and Nelson, R. (1999) Multipurpose trees as improved fallow: 

an economic assessment. The International Tree Crops Journal 10, 19-36. 
Groot, JJ.R. and Houba, V.J.G. (1995) A comparison of different indices for 

nitrogen mineralization. Biology and Fertility of Soils 19, 1-9. 
Grossmann, J. and Udluft, P. (1991) The extraction of soil water by the suction 

cup method: a review . Journal of Soil Science 42, 83-94. 
Guehl, J.M., Domenach, A.M., Bereau, M., Barigah, T.S., Casabianca, H., Ferhi, 

A. and Garbaye, J. (1998) Functional diversity in an Amazonian rainforest of 

French Guyana: a dual isotope approach (8 15 N and 8 13 C). Oecologia 116, 

Guggenberger, G., Haumaier, L., Thomas, R.J. and Zech, W. (1996) Assessing the 

organic phosphorus status of an oxisol under tropical pastures following native 

savannah using 31 P-NMR spectroscopy. Biology and Fertility of Soils 23, 332-339. 
Guindon, D. (1996) The importance of forest fragments to the maintenance of 

regional biodiversity in Costa Rica. In: Schelhas, J. and Greenberg, R. (eds) 

Forest Patches in Tropical Landscapes. Island Press, Washington, DC, pp. 

Gunjal, S.S. and Patil, P.L. (1992) Mycorrhizal control of wilt in casuarina. 

Agroforestry Today 4 (2), 14-15. 
Guo, Y., George, E. and Marschner, H. (1996) Contribution of an arbuscular 

372 Refer. 

mycorrhizal fungus to the uptake of cadmium and nickel in bean and maize 

plants. Plant and Soil 184, 195-205. 
Gupta, S.R. and Singh, J. S. (1981) The effect of plant species, weather variables 

and chemical composition of plant material on decomposition in a tropical 

grassland. Plant and Soil 59, 99-1 17. 
Habte, M. and Manjunath, A. (1987) Soil solution phosphorus status and 

mycorrhizal dependency in Leucaena leucocephala. Applied and Environmental 

Microbiology 53, 797-801. 
Haby, V.A., Russelle, M.P. and Skogley, E.O. (1990) Testing soils for potassium, 

calcium, and magnesium. In: Westerman, R.L. (ed.) Soil Testing and Plant 

Analysis. Soil Science Society of America, Madison, Wisconsin, pp. 181-227. 
Hagedorn, F., Steiner, K.G., Sekayange, L. and Zech, W. (1997) Effect of rainfall 

pattern on nitrogen mineralization and leaching in a green manure 

experiment in South Rwanda. Plant and Soil 195, 365-375. 
Hagen, M., Puhler, A. and Selbitschka, W. (1997) The persistence of 

bioluminescent Rhizobium meliloti strains LI (RecA(-)) and L33 (RecA( + )) in 

non-sterile microcosms depends on the soil type, on the co-cultivation of the 

host legume alfalfa and on the presence of an indigenous R. meliloti population. 

Plant and Soil 188, 257-266. 
Haggar, J.P., Warren, G.P., Beer, J. W. and Kass, D. (1991) Phosphorus availability 

under alley cropping and mulched and unmulched sole cropping systems in 

Costa Rica. Plant and Soil 137, 275-283. 
Haggar, J. P., Tanner, E.V.J., Beer, J.W. and Kass, D.C.L. (1993) Nitrogen 

dynamics of tropical agroforestry and annual cropping systems. Soil Biology 

and Biochemistry 25, 1363-1378. 
Hahn, A., Gobel, C. and Hock, B. (1998) Polyclonal antibodies for the detection 

of arbuscular mycorrhizal fungi. In: Varma, A. (ed.) Mycorrhiza Manual. 

Springer, Berlin, pp. 255-270. 
Haines, P.J. and Uren, N.C. (1990) Effects of conservation tillage farming on soil 

microbial biomass, organic matter and earthworm populations, in North- 
eastern Victoria. Australian Journal of Experimental Agriculture 30, 365-371. 
Hallaire, V. (1994) Description of microcrack orientation in a clayey soil using 

image analysis. Development in Soil Science 22, 549-557. 
Hallaire, V. and Curmi, P. (1994) Image analysis of pore space morphology in soil 

sections in relation to water movement. Development in Soil Science 22, 559-567. 
Halvorson, J.J., Smith, J. L., Bolton, H. Jr and Rossi, R.E. (1995) Evaluating shrub- 
associated patterns of soil properties in a shrub-steppe ecosystem using 

multiple-variable geostatistics. Soil Science Society of America Journal 59, 

Hamburg, S.P. (2000) Simple rules for measuring changes in ecosystem carbon in 

forestry-offset projects. Mitigation and Adaptation Strategies for Global Change 5, 

Hamel, C, Barrantes-Cartin, U., Furlan, V. and Smith, D.L. (1991) 

Endomycorrhizal fungi in nitrogen transfer from soybean to maize. Plant and 

Soil 138, 33-40. 
Handayanto, E., Cadisch, G. and Giller, K.E. (1994) Nitrogen release from 

prunings of legume hedgerow trees in relation to quality of the prunings and 

incubation method. Plant and Soil 160, 237-248. 

References 373 

Handayanto, E., Giller, K.E. and Cadisch, G. (1997) Nitrogen mineralization from 

mixtures of legume tree prunings of different quality and recovery of nitrogen 

by maize. Soil Biology and Biochemistry 29, 1417-1426. 
Handley, L.L., Odee, D. and Scrimgeour, CM. (1994) 5 15 N and 8 13 C patterns in 

savanna vegetation: dependence on water availability and disturbance. 

Functional Ecology 8, 306-314. 
Handley, L.L., Brendel, O., Scrimgeour, CM., Schmidt, S., Raven, J.A., Turnbull, 

M.H. and Stewart, G.R. (1996) The 15 N natural abundance patterns of field- 
collected fungi from three kinds of ecosystems. Rapid Communications in Mass 

Spectrometry 10, 974-978. 
Handley, L.L., Azcon, R., Ruiz Lozano, J.M. and Scrimgeour, CM. (1999) Plant 

8 15 N associated with arbuscular mycorrhization, drought and nitrogen 

deficiency. Rapid Communications in Mass Spectrometry 13, 1320-1324. 
Hanne, C. (2001) Die Rolle der Termiten im Kohlenstoffkreislauf eines 

amazonischen Festlandregenwaldes. PhD thesis, University of Frankfurt/Main, 

Frankfurt/Main, 156 pp. 
Hanson, P.J., Edwards, N.T., Garten, C.T. and Andrews, J.A. (2000) Separating 

root and soil microbial contributions to soil respiration: a review of methods 

and observations. Biogeochemistry 48, 115-146. 
Hansson, A.-C and Andren, O. (1986) Below-ground plant production in a 

perennial grass ley (Festuca pratensis Huds.) assessed with different methods. 

Journal of Applied Ecology 23, 657-666. 
Hardie, K. (1985) The effect of removal of extraradical hyphae on water uptake 

by vesicular-arbuscular mycorrhizal plants. New Phytologist 101, 677-684. 
Harlow, E. and Lane, D. (1988) Antibodies. A Laboratory Manual. Cold Spring Harbor 

Laboratory Press, Cold Spring Harbor, New York, 726 pp. 
Harper, S.M., Edwards, D.G., Kerven, G.L. and Asher, C.J. (1995) Effects of 

organic acid fractions extracted from Eucalyptus camaldulensis leaves on root 

elongation of maize (Tea mays) in the presence and absence of aluminium. Plant 

and Soil 171, 189-192. 
Hart, S.C, Stark, J.M., Davidson, E.A. and Firestone, M.K. (1994) Nitrogen 

mineralization, immobilization, and nitrification. In: Weaver, R.W., Angle, S., 

Bottomley, P., Bezdicek, D., Smith, S., Tabatabai, A. and Wollum, A. (eds) 

Methods of Soil Analysis, Part 2. Soil Science Society of America, Madison, 

Wisconsin, pp. 985-1018. 
Hartemink, A.E., Buresh, R.J. , Jama, B. and Janssen, B.H. (1996) Soil nitrate and 

water dynamics in Sesbania fallows, weed fallows, and maize. Soil Science Society 

of America Journal 60, 568-574. 
Hartemink, A.E., Buresh, R.J., van Bodegom, P.M., Braun, A.R., Jama, B. and 

Janssen, B.H. (2000) Inorganic nitrogen dynamics in fallows and maize on an 

Oxisol and Alfisol in the highlands of Kenya. Geoderma 98, 11-33. 
Haselwandter, K. and Bowen, G.D. (1996) Mycorrhizal relations in trees for 

agroforestry and land rehabilitation. Forest Ecology and Management 81, 

Hassal, M. and Rushton, S.P. (1982) The role of coprophagy in the feeding 

strategies of terrestrial isopods. Oecologia 53, 374-381. 
Hassink, J. (1995) Density fractions of macroorganic matter and microbial biomass as 

predictors of C and N mineralization. Soil Biology and Biochemistry 27, 1099-1 108. 

374 Refer. 

Hassink,J., Neutel, A. and de Ruiter, P.C. (1994) C and N mineralization in sandy 

and loamy soils: the role of microbes and microfauna. Soil Biology and 

Biochemistry 26, 1565-1571. 
Haumaier, L. and Zech, W. (1995) Black carbon - possible source of highly 

aromatic components of soil humic acids. Organic Geochemistry 23, 191-196. 
Hauser, S. (1993) Root distribution of Dactyladenia (Acioa) barteri and Senna (Cassia) 

siamea in alley cropping on Ultisol. I. Implication for field experimentation. 

Agroforestry Systems 24, 111-121. 
Hauser, S. and Gichuru, M.P. (1994) Root distribution of Dactyladenia (Acioa) barteri 

and Senna (Cassia) siamea in alley cropping on Ultisol II. Impact on water 

regime and consequences for experimental design. Agroforestry Systems 26, 

Havlin, J.L., Beaton, J.D., Tisdale, S.L. and Nelson, W.L. (1999) Soil Fertility and 

Fertilizers - an Introduction to Nutrient Management. Prentice Hall, Englewood 

Cliffs, New Jersey, 499 pp. 
Haynes, R.J. and Swift, R.S. (1990) Stability of soil aggregates in relation to organic 

constituents and soil water content. Journal of Soil Science 41, 73-83. 
He, Z.L. and Zhu, J. (1997) Transformation and bioavailability of specifically sorbed 

phosphate on variable-charge minerals in soils. Biology and Fertility of Soils 25, 

Heal, O.W., Anderson, J.M. and Swift, M.J. (1997) Plant litter quality and 

decomposition: an historical overview. In: Cadisch, G. and Giller, K.E. (eds) 

Driven by Nature: Plant Litter Quality and Decomposition. CAB International, 

Wallingford, UK, pp. 3-30. 
Hedley, M.J., Stewart, J. W.B. and Chauhan, R.S. (1982) Changes in inorganic and 

organic phosphorus fractions induced by cultivation practices and by 

laboratory incubations. Soil Science Society of America Journal 52, 1593-1596. 
Heitefuss, R. (1987) Pflanzenschutz. Georg Thieme, Stuttgart, 342 pp. 
Helgason, T., Fitter, A.H. and Young, J.P.W. (1999) Molecular diversity of 

arbuscular mycorrhizal fungi colonising Hyanthoides non-scripta (bluebell) in a 

seminatural woodland. Molecular Ecology 8, 659-666. 
Hellin, J. and Larrea, S. (1997) Live barriers on hillside farms: are we really 

addressing farmers' needs? Agroforestry Forum 8, 17-21. 
Hendershot, W.H., Lalande, H. and Duquette, M. (1993a) Ion exchange and 

exchangeable cations. In: Carter, M.R. (ed.) Soil Sampling and Methods of 

Analysis. Lewis Publishers, Boca Raton, Florida, pp. 167-176. 
Hendershot, W.H., Lalande, H. and Duquette, M. (1993b) Soil reaction and 

exchangeable acidity. In: Carter, M.R. (ed.) Soil Sampling and Methods of 

Analysis. Lewis Publishers, Boca Raton, Florida, pp. 141-145. 
Hendrick, R.L. and Pregitzer, K.S. (1996) Applications of minirhizotrons to 

understand root function in forests and other natural ecosystems. Plant and 

Soil 185, 293-304. 
Hendrickx, J.M.H., Wierenga, P.J. and Nash, M.S. (1990) Variability of soil water 

tension and soil water content. Agricultural Water Management 18, 135-148. 
Hendrickx, J.M. H., Nieber, J.L. and Siccaman, P.D. (1994) Effect of tensiometer 

cup size on field soil water tension variability. Soil Science Society of America 

Journal 58, 309-315. 
Heneghan, L., Coleman, D.C., Zou, X., Crossley, D.A. Jr and Haines, B.L. (1999) 

References 375 

Soil microarthropod contributions to decomposition dynamics: tropical- 
temperate comparisons of a single substrate. Ecology 80, 1873-1882. 
Henin, S., Monnier, G. and Combeau, A. (1958) Methode pour l'etude de la 

stabilite structurale des sols. Annates Agronomiques 9, 73-92. 
Hepper, CM. (1977) A colorimetric method for estimating vesicular-arbuscular 

mycorrhizal infection in roots. Soil Biology and Biochemistry 9, 15-18. 
Herkelrath, W.N., Hamburg, S.P. and Murphy, F. (1991) Automatic, real-time 

monitoring of soil moisture in a remote field area with time domain 

reflectometry. Water Resources Research 27, 857-864. 
Hernandez Lucas, I., Segovia, L., Martinez-Romero, E. and Pueppke, S.G. (1995) 

Phylogenetic relationships and host range of Rhizobium spp. that nodulate 

Phaseolus vulgaris L. Applied and Environmental Microbiology 61, 2775-2779. 
Herridge, D.F., Palmer, B., Nurhayati, D.P. and Peoples, M.B. (1996) Evaluation 

of the xylem ureide method for measuring N 2 fixation in six tree legume 

species. Soil Biology and Biochemistry 28, 281-289. 
Heuer, H., Hartung, K., Wieland, G., Kramer, K. and Smalla, K. (1999) 

Polynucleotide probes that target a hypervariable region of 16S rRNA genes 

to identify bacterial isolates corresponding to bands of community 

fingerprints. Applied and Environmental Microbiology 65, 1045-1049. 
Hillel, D. (1998) Environmental Soil Physics. Academic Press, San Diego, California, 

771 pp. 
Hindell, R.P., McKenzie, B.M. and Tisdall, J.M. (1997) Influence of drying and 

ageing on the stabilization of earthworm (Lumbricidae) casts. Biology and 

Fertility of Soils 25, 27-35. 
Hobbie, E.A., Macko, S.A. and Shugart, H.H. (1999) Interpretation of nitrogen 

isotope signatures using the NIFTE model. Oecologia 120, 405-415. 
Hobbie, E.A., Macko, S.A. and Williams, M.K. (2000) Correlations between foliar 

delta 15 N and nitrogen concentrations may indicate plant-mycorrhizal 

interactions. Oecologia 122, 273-283. 
Hoben, Q.A. (1995) Paradigms and politics: the cultural construction of 

environmental policy in Ethiopia. World Development 23, 1007-1022. 
Hodge, A., Paterson, E., Grayston, S.J., Campbell, CD., Ord, B.G. and Killham, 

K. (1998) Characterisation and microbial utilisation of exudate material from 

the rhizosphere of Lolium perenne grown under C0 2 enrichment. Soil Biology 

and Biochemistry 30,1033-1043. 
Hogberg, P. (1989) Root symbioses of trees in savannas. In: Proctor, J. (ed.) Mineral 

Nutrients in Tropical Forest and Savanna Ecosystems. Blackwell Scientific 

Publications, Oxford, pp. 121-135. 
Hogberg, P. (1990) lo N natural abundance as possible marker of the 

ectomycorrhizal habit of trees in a mixed African woodland. New Phytologist 

Hogberg, P. and Alexander, I.J. (1995) Roles of root symbioses in African woodland 

and forest: evidence from 15 N abundance and foliar analysis. Journal of Ecology 

Hojberg, O., Schnider, U., Winteler, H.V., Sorensen, J. and Haas, D. (1999) 

Oxygen-sensing reporter strain of Pseudomonas ftuorescens for monitoring the 

distribution of low-oxygen habitats in soil. Applied and Environmental 

Microbiology 65, 4085-4093. 

376 Refer. 

Holford, I.C.R. (1979) Evaluation of soil phosphate buffering capacity indices. 

Australian Journal of Soil Research IV, 495-504. 
Holland, E.A. and Coleman, D.C. (1987) Litter placement effects on microbial and 

organic matter dynamics in an agroecosystem. Ecology 68, 425-433. 
Hooper, D.U., Bignell, D.E., Brown, V.K., Brussaard, L., Dangerfield, J.M., Wall, 

D., Wardle, D.A., Coleman, D.C, Ciller, K.E., Lavelle, P., van der Putten, 

W.H., de Ruiter, P., Rusek, J., Sala, O., Silver, W., Tiedje, J. and Wolters, V. 

(2000) Interactions between above and belowground biodiversity in terrestrial 

ecosystems: patterns, mechanisms and feedbacks. BioScience 50, 1049-1061. 
Hoosbeck, M.R. and Bouma, J. (1998) Obtaining soil and land quality indicators 

using research chains and geostatistical methods. Nutrient Cycling in 

Agroecosystems 50, 35-50. 
Hornung, M. (1990) Measurement of nutrient losses in soil erosion. In: Harrison, 

A.F., Ineson, P. and Heal, O.W. (eds) Nutrient Cycling in Terrestrial Ecosystems. 

Elsevier, London, pp. 80-102. 
Horst, W.J. (1995) Efficiency of soil-nutrient use in intercropping systems. In: 

Sinquet, A. and Cruz, P. (eds) Ecophysiology of Tropical Intercropping. Institut 

National de Recherche Agronomique, Paris, pp. 197-211. 
House, G.J. (1985) Comparison of soil arthropods and earthworms from 

conventional and no-tillage agroecosystems. Soil and Tillage Research 5, 

van Hove, L.W.A., Adema, E.H., Vredenberg, W.J. and Pieters, G.A. (1989) A study 

of the adsorption of NH 3 and S0 2 on leaf surfaces. Atmospheric Environment 23, 

Howard, PJ.A. (1997) Analysis of data from BIOLOG plates: comments on the 

method of Garland and Mills. Soil Biology and Biochemistry 29, 1755-1757. 
Hudson, N.W. (1992) Land Husbandry. Batsford, London, 192 pp. 
Hudson, N.W. (1995) Soil Conservation. Batsford, London, 391 pp. 
Hue, N.V. (1991) Effects of organic acids/anions on P sorption and phytoa variability 

in soils with different mineralogies. Soil Science 152, 463-471. 
Hughes, R.F., Kauffman, J.B. and Cummings, D.L. (2000) Fire in the Brazilian 

Amazon: 3. Dynamics of biomass, C, and nutrient pools in regenerating 

forests. Oecologia 124, 574-588. 
Huguet, J.G. (1973) A new method of studying the rooting of perennial plants by 

means of a spiral trench. Annals of Agronomy 24, 707-73 1. 
Hulugalle, N.R. (1994) Long-term effects of land clearing methods, tillage systems 

and cropping systems on surface soil properties of a tropical Alfisol in S.W. 

Nigeria. Soil Use and Management 10, 25-30. 
Hunt, HW. (1987) The detrital food web in a shortgrass prairie. Biology and Fertility 

of Soils 3, 57-68. 
Hurni, H. and Nuntapong, S. (1983) Agro-forestry improvement for shifting 

cultivation systems: soil conservation research in Northern Thailand. Mountain 

Research and Development 3, 338-345. 
Huwe, B. (1993) Geostatistische Untersuchungen zum Wasser- und 

Stickstoffhaushalt landwirtschaftlich genutzter Boden im Einzugsgebiet eines 

Wasserwerks. VDLUFA Schriftenreihe 37, 145-148. 
Huxley, P. (1999) Tropical Agroforestry. Blackwell Science, Oxford, 371 pp. 
Huxley, P.A., Patel, R.Z., Kabaara, A.M. and Mitchell, H.W. (1974) Tracer studies 

References 377 

with 32 P on the distribution of functional roots of Arabica coffee in Kenya. 

Annals of Applied Biology 77, 159-180. 
Huxley, P.A., Pinney, A., Akunda, E. and Muraya, P. (1994) A tree/crop interface 

orientation experiment with a Grevillea robusta hedgerow and maize. 

Agroforestry Systems 26, 23-45. 
IAEA (1975) Root Activity Patterns of some Tree Crops. Technical Report Series No. 

170. International Atomic Energy Agency, Vienna, 154 pp. 
Ikerra, T.W.D., Mneki, P.N.S. and Singh, B.R. (1994) Effects of added compost 

and farmyard manure on P release from Minjingo phosphate rock and its 

uptake by maize. Norwegian Journal of Agricultural Science 8, 13-23. 
Ikram, A., Jensen, E.S. and Jakobsen, I. (1994) No significant transfer of N and P 

from Pueraria phaseoloides to Hevea brasiliensis via hyphal links of arbuscular 

mycorrhiza. Soil Biology and Biochemistry 26, 1541-1547. 
Imbach, A.C., Fassbender, H.W., Borel, R., Beer, J. and Bonnemann, A. (1989) 

Modelling agroforestry systems of cacao (Theobraoma cacao) with laurel (Cordia 

alliodora) and cacao with poro (Erythrina poeppigiana) in Costa Rica 4. Water 

balances, nutrient inputs and leaching. Agroforestry Systems 8, 267-287. 
Innocenti, G. and Branzanti, B. (1998) Interactions between AM fungi and soil- 
borne cereal pathogen fungi. In: Abstracts, Second International Conference on 

Mycorrhizas. Swedish University of Agricultural Sciences, Uppsala, p. 86. 
Irwin, H.S. and Barneby, R.C. (1981) Cassieae. In: Polhill, R.M. and Raven, P.H. 

(eds) Advances in Legume Systematics 1. Royal Botanic Gardens, Kew, pp. 97-106. 
Ives, J. and Masserli, B. (1989) The Himalayan Dilemma: Reconciling Development and 

Conservation. Routledge, London, 295 pp. 
Iwata, S., Tabuchi, T. and Warkentin, B.P. (1995) Soil-Water Interactions. Marcel 

Dekker, New York, 440 pp. 
Iyamuremye, F. and Dick, R.P. (1996) Organic amendments and phosphorus 

sorption by soils. Advances in Agronomy 56, 139-185. 
Iyamuremye, F., Dick, R.P. and Baham, J. (1996) Organic amendments and 

phosphorus dynamics: II. Distribution of soil phosphorus fractions. Soil Science 

Izac, A.-M.N. (1997) Ecological economics of investing in natural capital in Africa. 

In: Buresh, R.J., Sanchez, PA. and Calhoun, F. (eds) Replenishing Soil Fertility 

in Africa. Soil Science Society of America, Madison, Wisconsin, pp. 237-251. 
Izac, A.-M.N. and Sanchez, PA. (2001) Towards a natural resource management 

paradigm for international agriculture: the example of agroforestry research. 

Agricultural Systems 69, 5-25. 
Jackson, P.C., Cavelier, J., Goldstein, G., Meinzer, F.C. and Holbrook, N.M. (1995) 

Partitioning of water resources among plants of a lowland tropical forest. 

Oecologia 101, 197-203. 
Jackson, P.C., Meinzer, F.C, Bustamante, M., Goldstein, G., Franco, A., Rundel, 

P.W., Caldas, L., Igler, E. and Causin, F. (1999) Partitioning of soil water 

among tree species in a Brazilian Cerrado ecosystem. Tree Physiology 19, 

Jaeger, C.H., Lindow, S.E., Miller, S., Clark, E. and Firestone, M.K. (1999) 

Mapping of sugar and amino acid availability in soil around roots with 

bacterial sensors of sucrose and tryptophan. Applied and Environmental 

Microbiology 65, 2685-2690. 

378 Refer. 

Jama, B., Swinkels, R.A. and Buresh, R.J. (1997) Agronomic and economic 

evaluation of organic and inorganic sources of phosphorus in Western Kenya. 

Agronomy Journal 89, 597-604. 
Jama, B., Buresh, R.J., Ndufa, J.K. and Shepherd, K.D. (1998) Vertical distribution 

of roots and soil nitrate: tree species and phosphorus effects. Soil Science Society 

of America Journal 62, 280-286. 
Jama, B., Palm, C.A., Buresh, R.J., Niang, A., Gachengo, C, Nziguheba, G. and 

Amadalo, B. (2000) Tithonia diversifolia as a green manure for soil fertility 

improvement in western Kenya: a review. Agroforestry Systems 49, 201-221. 
James, D.W. and Wells, K.L. (1990) Soil sample collection and handling: technique 

based on source and degree of field variability. In: Westerman, R.L. (ed.) Soil 

Testing and Plant Analysis. Soil Science Society of America, Madison, Wisconsin, 

pp. 25-44. 
Janzen, H.H. and Ellert, B.H. (1998) Sulfur dynamics in cultivated temperate 

agroecosystems. In: Maynard, D.G. (ed.) Sulfur in the Environment. Marcel 

Dekker, New York, pp. 11-43. 
Janzen, H.H. and McGinn, S.M. (1991) Volatile loss of nitrogen during decompo- 
sition of legume green manure. Soil Biology and Biochemistry 23, 291-297. 
Jarstfer, A.G. and Sylvia, D.M. (1995) Aeroponic culture of VAM fungi. In: Hock, 

B. and Varma, A. (eds) Mycorrhiza - Structure, Function, Molecular Biology and 

Biotechnology. Springer, Berlin, pp. 427-442. 
Jastrow, J.D. and Miller, R.M. (1991) Methods for assessing the effects of biota on 

soil structure. Agriculture, Ecosystems and Environment 34, 279-303. 
Jegou, D., Hallaire, V., Cluzeau, D. and Trehen, P. (1999) Characterization of the 

burrow system of the earthworms Lumbricus terrestris and Aporectodea giardi 

using X-ray computed tomography and image analysis. Biology and Fertility of 

Soils 29, 314-318. 
Jenkinson, D.S. (1988a) Determination of microbial biomass carbon and nitrogen 

in soil. In: Wilson, J.R. (ed.) Advances in Nitrogen Cycling in Agricultural 

Ecosystems. CAB International, Wallingford, UK, pp. 368-386. 
Jenkinson, D.S. (1988b) Soil organic matter and its dynamics. In: Wild, A. (ed.) 

Russell's Soil Conditions and Plant Growth. Longman, Harlow, pp. 564-607. 
Jenkinson, D.S. (1990) The turnover of organic carbon and nitrogen in 

soil. Philosophical Transactions of the Royal Society of London, Series B 329, 

Jenkinson, D.S. and Ayanaba, A. (1977) Decomposition of carbon-14 labeled plant 

material under tropical conditions. Soil Science Society of America Journal 41, 

Jenkinson, D.S., Fox, R.H. and Rayner, J.H. (1985) Interactions between fertilizer 

nitrogen and soil nitrogen - the so-called "priming effect" . Journal of Soil 

Science 36, 425-444. 
Jensen, L.E., Kragelund, L. and Nybroe, O. (1998) Expression of a nitrogen 

regulated lux gene fusion in Pseudomonas fluorescens DF57 studied in pure 

culture and in soil. FEMS Microbiology Ecology 25, 23-32. 
Jensen, L.S. and Sorensen, J. (1994) Microscale fumigation-extraction and 

substrate-induced respiration methods for measuring microbial biomass in 

barley rhizosphere. Plant and Soil 162, 151-161. 
Jentschke, G., Brandes, B., Kuhn, A.J., Schroder, W.H., Becker, J. S. and Godbold, 

References 379 

D.L. (2000) The mycorrhizal fungus Paxillus involutus transports magnesium 

to Norway spruce seedlings. Evidence from stable isotope labeling. Plant and 

Soil 220, 243-246. 
Johansen, A. and Jensen, E.S. (1996) Transfer of N and P from intact or 

decomposing roots of pea or barley interconnected by an arbuscular 

mycorrhizal fungus. Soil Biology and Biochemistry 28, 73-81. 
Johnson, D.L. (1990) Biomantle evolution and the redistribution of earth materials 

and artifacts. Soil Science 149, 84-102. 
Johnson, M.G., Tingey, D.T., Phillips, D.L. and Storm, M.J. (2001) Advancing fine 

root research with minirhizotrons. Environmental and Experimental Botany 45, 

Johnson, N.C. and Pfleger, F.L. (1992) Vesicular-arbuscular mycorrhizae and 

cultural stresses. In: Bethlenfalvay, G.J. and Linderman, R.G. (eds) Mycorrhizae 

in Sustainable Agriculture . American Society of Agronomy, Madison, Wisconsin, 

pp. 71-99. 
Joner, E.J., Magid, J., Gahoonia, T.S. and Jakobsen, LP. (1995) Depletion and 

activity of phosphatases in the rhizosphere of mycorrhizal and nonmycorrhizal 

cucumber (Cucumis sativus L.). Soil Biology and Biochemistry 27, 1 145-1 15 1. 
Jones, D.L. (1998) Organic acids in the rhizosphere - a critical review. Plant and 

Soil 205, 25-44. 
Jones, D.L. and Darrah, P.R. (1993) Re-sorption of organic compounds by roots 

oiZea mays L. and its consequences in the rhizosphere. 2. Experimental and 

model evidence for simultaneous exudation and re-sorption of soluble C 

compounds. Plant and Soil 153, 47-59. 
Jones, D.L. and Darrah, P.R. (1994) Simple method for 14 C-labelling root material 

for use in root decomposition studies. Communications in Soil Science and Plant 

Analysis 25, 2737-2743. 
Jones, D.L. and Edwards, A.C. (1993) Effect of moisture content and preparation 

technique on the composition of soil solution obtained by centrifugation. 

Communications in Soil Science and Plant Analysis 24, 171-186. 
Jones, D.T. and Eggleton, P. (2000) Sampling termite species assemblages in 

tropical forests: testing a rapid biodiversity assessment protocol. Journal of 

Applied Ecology 37, 191-203. 
Jones, M., Sinclair, F.L. and Grime, V.L. (1998) Effect of tree species and crown 

pruning on root length and soil water content in semi-arid agroforestry. Plant 

and Soil 201, 197-207. 
Jongmans, A.G., van Breemen, N., Lundstrom, U., Finlay, R.D., van Hees, P.A.W., 

Giesler, R., Melkerud, PA., Olsson, M., Srinivasan, M. and Unestam, T. (1997) 

Rock eating fungi. Nature 389, 682-683. 
Jonsson, K., Fidjeland, L., Maghembe, J A. and Hogberg, P. (1988) The vertical 

distribution of fine roots of five tree species and maize in Morogoro, Tanzania. 

Agroforestry Systems 6, 63-69. 
Jonsson, K., Ong, C.K. and Odongo, J.C.W. (1999) Influence of scattered nere 

and karite trees on microclimate, soil fertility and millet yield in Burkina Faso. 

Experimental Agriculture 35, 39-53. 
Jordan, C.F. (1968) A simple, tension-free lysimeter. Soil Science 105, 81. 
Jordan, C.F. and Escalante, G. (1980) Root productivity in an Amazonian rain 

forest. Ecology 61, 14-18. 


Jorge, L.A.C. (1999) Descricao detalhada do metodo de trincheira com producao 

de imagens para uso do SIARCS®. In: Fernandes, M.F., Tavares, E.D. and 

da Silva Leal, M.L. (eds) Workshop sobre Sistema Radicular: Metodologias e Estudo 

de Casos. 27-29 Oct. 1999, Aracaju . Embrapa Tabuleiros Costeiros, Aracaju, 

pp. 255-268. 
Jose, S., Gillespie, A.R., Seifert, J.R. and Biehle, D.J. (2000) Denning competition 

vectors in a temperate alley cropping system in the midwestern USA. 2. 

Competition for water. Agroforestry Systems 48, 41-59. 
Jose, S., Gillespie, A.R., Seifert, J.R. and Pope, P.E. (2001) Comparison of 

minirhizotron and soil core methods for quantifying root biomass in a 

temperate alley cropping system. Agroforestry Systems 52, 161-168. 
Josens, G. (1983) The soil fauna of tropical savannas III. The termites. In: 

Bourliere, F. (ed.) Tropical Savannas. Elsevier, New York, pp. 498-513. 
Journel, A.G. (1989) Fundamentals of Geostatistics in Five Lessons. American 

Geophysical Union, Washington, DC, 40 pp. 
Journel, A.G. and Huijbregts, C.J. (1978) Mining Geostatistics. Academic Press, 

London, 599 pp. 
Juo, A.S.R. and Lai, R. (1977) The effect of fallow and continuous cultivation on 

the chemical and physical properties of an alfisol in western Nigeria. Plant and 

Soil 47, 567-584. 
Juo, A.S.R., Franzluebbers, K., Dabiri, A. and Ikhile, B. (1995) Changes in soil 

properties during long-term fallow and continuous cultivation after forest 

clearing in Nigeria. Agriculture, Ecosystems and Environment 56, 9-18. 
Jury, W.A., Gardner, W.R. and Gardner, W.H. (1991) Soil Physics. John Wiley & 

Sons, New York, 328 pp. 
Kamara, A.Y., Akobundu, I.O., Sanginga, N. and Jutzi, S.C. (1999) Effects of mulch 

from 14 multipurpose tree species (MPTs) on early growth and nodulation of 

cowpea (Vigna unguiculata L.). Journal of Agronomy and Crop Science 182, 

Kamh, M., Horst, W.J., Amer, F., Mostafa, H. and Maier, P. (1999) Mobilization 

of soil and fertilizer phosphate by cover crops. Plant and Soil 21 1, 19-27. 
Kanbur, R. and Squire, L. (2000) The evolution of thinking about poverty: 

exploring the interactions. In: World Development Report on Poverty, Background 

Papers. Available at: 
Kang, B.T., Okoro, E., Acquaye, D.F. and Osiname, OA. (1981) Sulfur status of 

some Nigerian soils from the savanna and forest zone. Soil Science 132, 220-227. 
Kang, B.T., Grimme, H. and Lawson, T.L. (1985) Alley cropping sequentially 

cropped maize and cowpea with leucaena on a sandy soil in southern Nigeria. 

Plant and Soil 85, 267-276. 
Kangas, P. (1992) Root regrowth in a subtropical wet forest in Puerto Rico. 

Biotropica 24, 463-465. 
Katterer, T., Fabiao, A., Madeira, M., Ribeiro, C. and Steen, E. (1995) Fine-root 

dynamics, soil moisture and soil carbon content in a Eucalyptus globulus 

plantation under different irrigation and fertilisation regimes. Forest Ecology 

and Management 74, 1-12. 
Kauffman, J.B., Cummings, D.L. and Ward, D.E. (1998) Fire in the Brazilian 

Amazon 2. Biomass, nutrient pools and losses in cattle pastures. Oecologia 1 13, 


References 381 

Keeney, D.R. (1982) Nitrogen - availability indices. In: Page, A.L., Miller, R.H. 

and Keeney, D.R. (eds) Methods of Soil Analysis, Part 2. American Society of 

Agronomy, Madison, Wisconsin, pp. 711-733. 
Kellman, M. (1979) Soil enrichment by neotropical savanna trees. Journal of Ecology 

67, 565-577. 
Kellman, M. and Carty, A. (1986) Magnitude of nutrient influxes from atmospheric 

sources to a central American Pinus caribaea woodland. Journal of Applied 

Ecology 23, 211-226. 
Kerven, G.L., Larsen, P.L., Bell, L.C. and Edwards, D.G. (1995) Quantitative 27 A1 

NMR spectroscopic studies of Al (III) complexes with organic acid ligands 

and their comparison with GEOCHEM predicted values. Plant and Soil 171, 

Kessler, J.J. and Breman, H. (1991) The potential of agroforestry to increase 

primary production in the Sahelian and Sudanian zones of West Africa. 

Agroforestry Systems 13, 41-62. 
Khiewtam, R.S. and Ramakrishnan, P.S. (1993) Litter and fine root dynamics of a 

relict sacred grove forest at Cherrapunji in north-eastern India. Forest Ecology 

and Management 60, 327-344. 
Kiepe, P. (1995) Effect of Cassia siamea hedgerow barriers on soil physical 

properties. Geoderma 66, 113-120. 
Kim, D.J., Vanclooster, M., Feyen, J. and Vereecken, H. (1998) Simple linear 

model for calibration of time domain reflectometry measurements on solute 

concentration. Soil Science Society of America Journal 62, 83-89. 
Kinraide, T.B. (1991) Identity of the rhizotoxic aluminium species. Plant and Soil 

134, 167-178. 
Kirk, G.J.D. (1999) A model of phosphate solubilization by organic anion excretion 

from plant roots. European Journal of Soil Science 50, 369-378. 
Kirk, G.J.D. and Solivas, J.L. (1994) Coupled diffusion and oxidation of ferrous 

iron in soils. 3. Further development of the model and experimental testing. 

European Journal of Soil Science 45, 369-378. 
Kirk, G.J.D., Santos, E.E. and Findenegg, G.R. (1999) Phosphate solubilization by 

organic anion excretion from rice (Oryza sativa L.) growing in aerobic soil. Plant 

and Soil 211, 11-18. 
Kitanidis, P.K. (1997) Introduction to Geostatistics, Applications in Hydrogeology. 

Cambridge University Press, New York, 249 pp. 
Klute, A. (1986) Methods of Soil Analysis, Part 1. American Society of Agronomy, 

Madison, Wisconsin. 
Koide, R. (1985) The effect of VA mycorrhizal infection and phosphorus status on 

sunflower hydraulic and stomata properties. Journal of Experimental Botany 36, 

de Koning, G.H.J., van de Kop, P.J. and Fresco, L.O. (1997) Estimates of sub- 
national nutrient balances as sustainability indicators for agro-ecosystems in 

Ecuador. Agriculture, Ecosystems and Environment 65, 127-139. 
Koske, R.E. and Gemma, J.N. (1989) A modified procedure for staining roots to 

detect VA mycorrhizas. Mycological Research 92, 486-505. 
Kowalenko, C.G. (1993a) Extraction of available sulfur. In: Carter, M.R. (ed.) Soil 

Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, Florida, pp. 


382 Refer. 

Kowalenko, C.G. (1993b) Total and fractions of sulfur. In: Carter, M.R. (ed.) Soil 

Sampling and Methods of Analysis. Lewis Publishers, Boca Raton, Florida, pp. 

Kowalenko, C.G. and Yu, S. (1996) Solution, exchangeable and clay-fixed 

ammonium in south coast British Columbia soils. Canadian Journal of Soil 

Science 76, 473-483. 
Kragelund, L., Hosbond, C. and Nybroe, O. (1997) Distribution of metabolic 

activity and phosphate starvation response of lux-tagged Pseudomonas 

fluorescens reporter bacteria in the barley rhizosphere. Applied and 

Environmental Microbiology 63, 4920-4928. 
Krouse, H.R., Stewart, J.W.B. and Grinenko, V.A. (1991) Pedosphere and 

biosphere. In: Krouse, H.R. and Grinenko, V.A. (eds) Stable Isotopes: Natural 

and Anthropogenic Sulphur in the Environment. John Wiley & Sons, New York, 

pp. 267-306. 
Kiihne, R.F. (1993) Wasser- und Nahrstoffhaushalt in Mais-Maniok-Anbausystemen 

mit und ohne Integration von Alleekulturen ("Alley cropping") in Sud-Benin. 

PhD thesis, University of Hohenheim, Hohenheim, Germany, 244 pp. 
Kummerow, J., Kummerow, M. and Souza da Silva, W. (1982) Fine-root growth 

dynamics of cacao (Theobroma cacao). Plant and Soil 65, 193-201. 
Kummerow, J., Castillanos, J., Maas, M. and Larigauderie, A. (1990) Production 

of fine roots and the seasonality of their growth in a Mexican deciduous dry 

forest. Vegetatio 90, 73-80. 
Kuono, K., Tuchiya, Y. and Ando, T. (1995) Measurement of soil microbial biomass 

phosphorus by an anion exchange membrane method. Soil Biology and 

Biochemistry 27, 1353-1357. 
Kurz, W.A. and Kimmins, J.P. (1987) Analysis of some sources of error in methods 

used to determine fine root production in forest ecosystems: a simulation 

approach. Canadian Journal of Forest Research 17, 909-912. 
Kutilek, M. and Nielsen, D.R. (1994) Soil Hydrology. Catena, Cremlingen-Destedt, 

370 pp. 
Kwesiga, F., Franzel, S., Place, F., Phiri, D. and Simwanza, C.P. (1999) Sesbania 

sesban improved fallows in eastern Zambia: their inception, development and 

farmer enthusiasm. Agroforestry Systems 47, 49-66. 
Ladd, J.N. (1981) The use of lo N in following organic matter turnover, with specific 

reference to rotation systems. Plant and Soil 58, 401-41 1. 
de Lajudie, P., Willems, A., Pot, B., Dewettinck, D., Maestrojuan, G., Neyra, M., 

Collins, M.D., Dreyfus, B., Kersters, K. and Gillis, M. (1994) Polyphasic 

taxonomy of rhizobia: emendation of the genus Sinorhizobium and description 

of Sinorhizobium meliloti comb, nov., Sinorhizobium sajeli sp. nov. and Sinorhizobium 

teranga sp. nov. International Journal of Systematic Bacteriology 44, 715-733. 
de Lajudie, P., Willems, A., Nick, G., Moreira, F., Molouba, F., Torek, U., Neyra, 

M., Collins, M.D., Lindstrom, K., Dreyfus, B. and Gillis, M. (1998) 

Characterization of tropical tree rhizobia and description of Mesorhizobium 

plurifarium sp. nov. International Journal of Systematic Bacteriology 48, 369-382. 
Lai, R. (1986) Soil surface management in the tropics for intensive land use and 

high and sustained production. Advances in Soil Science 5, 1-109. 
Lai, R. (1989) Agroforestry systems and soil surface management on a tropical 

alfisol. I. Soil moisture and crop yields. Agroforestry Systems 8, 7-29. 

References 383 

Lai, R. (ed.) (1994) Soil Erosion Research Methods. Soil and Water Conservation 

Society, Ankeny, 340 pp. 
Lamers, J.P.A. and Feil, P.R. (1995) Farmers' knowledge and management of 

spatial soil and crop growth variability in Niger, West Africa. Netherlands 

Journal of Agricultural Science 43, 375-389. 
Lamhamedi, M.S., Bernier, P.Y. and Fortin, JA. (1992) Hydraulic conductance 

and soil water potential at the soil root interface of Pinus pinaster seedlings 

inoculated with different dikaryons oiPisolithus sp. Tree Physiology 10, 231-244. 
Lamotte, M. (1989) Place des animaux detritivores et des microorganismes 

decomposeurs dans les flux d'energie des savanes africaines. Pedobiologia 33, 

Landeweert, R., Hoffland, E., Finlay, R.D., Kuyper, T.W. and van Breemen, N. 

(2001) Linking plants to rocks: ectomycorrhizal fungi mobilize nutrients from 

minerals. Trends in Ecology and Evolution 16, 248-254. 
Landon, J.R. (1991) Booker Tropical Soil Manual. Longman, Harlow, 474 pp. 
Lapeyrie, F., Chilvers, GA. and Behm, CA. (1987) Oxalic acid synthesis by the 

mycorrhizal fungus Paxillus involutus (Batsch ex Fr.) Fr. New Phytologist 106, 

Lascano, R.J. and Hatfield, J. L. (1992) Spatial variability of evaporation along two 

transects of a bare soil. Soil Science Society of America Journal 56, 341-346. 
Lassoie, J. P. and Hinckley, T.M. (1991) Techniques and Approaches in Forest Tree 

Ecophysiology. CRC Press, Boca Raton, Florida, 599 pp. 
Lavelle, P. (1975) Consommation annuelle de terre par une population naturelle 

de vers de terre (Millsonia anomala OMODEO, Acanthodrilidae, Oligochaeta) 

dans la savane de Lamto (Cote d'lvoire). Revue d'Ecologie et de Biologie du Sol 

12, 11-24. 
Lavelle, P. (1978) Les Vers de terre de la savane de Lamto (Cote d'lvoire): 

peuplements, populations et fonctions dans l'ecosysteme. PhD thesis, 

Universite Paris VI, Laboratoire de Zoologie de TENS, Paris, 301 pp. 
Lavelle, P. (1983) The soil fauna of tropical savannas II. The earthworms. In: 

Bourliere, F. (ed.) Tropical Savannas. Elsevier, New York, pp. 485-504. 
Lavelle, P. (1988) Assessing the abundance and role of invertebrate communities 

in tropical soils: aims and methods. Journal of African Zoology 102, 275-283. 
Lavelle, P. (1997) Faunal activities and soil processes: adaptive strategies that 

determine ecosystem function. Advances in Ecological Research 27, 93-132. 
Lavelle, P. (2000) Ecological challenges for soil science. Soil Science 165,